48860213-MEng Thesis Document Saikat Banerjee Final
Transcript of 48860213-MEng Thesis Document Saikat Banerjee Final
1
E-Commerce Based Closed-Loop Supply Chain for Plastic Recycling
By
Saikat Banerjee
Bachelor of Technology (B. Tech), Computer Science & Engineering West Bengal University of Technology (2010)
SUBMITTED TO THE PROGRAM IN SUPPLY CHAIN MANAGEMENT IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF
MASTER OF ENGINEERING IN SUPPLY CHAIN MANAGEMENT AT THE
MASSACHUSETTS INSTITUTE OF TECHNOLOGY (MIT) MAY 2020
© 2020 Saikat Banerjee. All Rights Reserved
The author hereby grants to MIT permission to reproduce and to distribute publicly paper and electronic copies of this thesis document in whole or in part in any medium now known or
hereafter created.
Signature of Author: ____________________________________________________________ Department of Supply Chain Management
May 2020
Certified by: ___________________________________________________________________ Dr. Eva Maria Ponce Cueto
Executive Director, MITx MicroMaster’s in Supply Chain Management Director, Omnichannel Distribution Strategies
Certified by: ___________________________________________________________________ Ms. Suzanne Greene
Program Manager, MIT Sustainable Supply Chains
Accepted by: __________________________________________________________________ Dr. Yossi Sheffi
Director, Center for Transportation and Logistics Elisha Gray II Professor of Engineering Systems Professor, Civil and Environmental Engineering
2
[This page is intentionally left blank]
3
E-Commerce Based Closed-Loop Supply Chain for Plastic Recycling
By
Saikat Banerjee
Submitted to The Program in Supply Chain Management on May 8, 2020 in Partial Fulfillment of the
Requirements for the Degree of Master of Engineering in Supply Chain Management
ABSTRACT
The world is facing a grave plastic waste problem. It is not new that we hear about oceanic death and morbid landfills. Only 8% of all the plastic produced is recycled in the US. This grotesque situation has been worsened by the Chinese ban of plastic waste imports from the developed western nations as of 2018. In this research we assess the feasibility of a novel approach to using existing e-commerce reverse logistics channels to take back post-consumer plastic. We use product sales data to estimate the post-consumer plastic volume. We then, design a mixed integer linear programming (MILP) based optimization model to assess different take-back routes and calculate various operational costs. In addition to the optimization model we determine the feasibility of this process by considering cost offsets such as price of virgin plastics. After that, we conduct a scenario-based sensitivity analysis to understand systemic cost and overall profit. We used the results of these analyses to formulate the strategic recommendations for companies interested in promoting or implementing e-commerce-based recycling programs. Finally, we assess the greenhouse gas emissions and corresponding externality costs through this process and perform a qualitative assessment of the stakeholder networks vital to making such a system operational. In conclusion, our results suggest that in certain scenarios it is economically feasible to facilitate a take-back process for post-consumer plastic using existing e-commerce-based reverse logistics channels while maintaining minimal additional emissions in the process. Thesis Advisor: Dr. Eva Maria Ponce Cueto Title: Executive Director, MITx MicroMasters in Supply Chain Management Director, OmniChannel Distribution Strategies Thesis Co-Advisor: Ms. Suzanne Greene Title: Program Manager, MIT Sustainable Supply Chains
4
[This page is intentionally left blank]
5
Acknowledgments
First and foremost, I would like to thank my thesis advisors, Dr. Eva Ponce and Ms.
Suzanne Greene, for their unwavering support, patience, and guidance. This thesis has been possible largely due to the time and resources they have invested in this work. Suzanne once told me, “I am pushing you to be the best”. I have always remembered that, and hope she feels the same after reading this paper. Eva has been my sounding board for my mathematical thought-process throughout this research endeavor. I would always be grateful to Eva and Suzanne. Thank you!
I am grateful to Dr. Tugba Efendigil for working with me to streamline the data collection process with respect to the location data and data related to the various systemic costs. Tugba has been a mentor and a friend throughout the process.
Thanks to my thesis committee members, Dr. Chris Caplice, Dr. Jarrod Goentzel, Dr. Josue Velazquez Martinez, and Dr. Maria Jesus Saenz, for their periodic feedback and suggestions to improve the output of my research.
In addition, I would like to thank Pamela Siska and Toby Gooley for reviewing the manuscript and providing valuable feedback. In Fall ’19, Pamela helped me articulate my thoughts better while I was composing the Introduction, Problem Statement and Literature Review sections of this paper. In Spring ’20, I benefited from the detail-oriented nature of Toby while reviewing this entire document. I am so grateful that I had an opportunity to work with Toby, without whom, the reader would be deprived of the pleasure, I would assume she would get from reading this paper.
Also, thanks to Justin Snow and Robert Cummings for all the administrative help during
the program. I would like to thank my parents, my father, Mr. Samir Kumar Banerjee, who introduced
me to Mathematics and encouraged me to take up challenges, making sure, I landed on softer ground if I failed; and my mother, Mrs. Runu Banerjee, who once told me, “If you do something, do it well, else don’t do it”. I will always remember that. Thank you for being a support system that I could constantly count on.
Finally, I would like to thank my wife, Ahana Roy Choudhury Banerjee for always being
a patient listener and an active compass from the initial ideation of this research to its completion, constantly supporting me in all ways possible. This work would not have been possible without her kindness and intellectual largess.
6
[This page is intentionally left blank]
7
Table of Contents
Table of Contents .......................................................................................................................................... 7 List of Figures ............................................................................................................................................... 8 List of Tables ................................................................................................................................................ 9 1. Introduction ......................................................................................................................................... 10 2. Problem Setting and Objectives .......................................................................................................... 15 3. Literature Review ................................................................................................................................ 17
3.1 Policies on Plastics ...................................................................................................................... 18 3.2 Consumer Response to Plastic Recycling and Take-Back Programs ......................................... 21 3.3 Recycling and the Potential Use of Collecting Post-Consumer Plastic ...................................... 21 3.4 Use of Reverse Logistics in Take-Back for Recycling ............................................................... 23 3.5 Use of E-Commerce in the Take-Back Process .......................................................................... 24 3.6 Aspects of Cost in the Take-Back Mechanisms .......................................................................... 26 3.7 Conclusion of Literature Review ................................................................................................ 27
4. Methodology ....................................................................................................................................... 29 4.1 Data Collection ........................................................................................................................... 29 4.2 Data Cleaning and Preparation ................................................................................................... 31 4.3 Initial Data Analysis .................................................................................................................... 33 4.4 Problem Formulation Using A Network Design Approach ........................................................ 35 4.5 Cost Analysis .............................................................................................................................. 38 4.6 Scenario-Based Sensitivity Analysis .......................................................................................... 40 4.7 Recommendations ....................................................................................................................... 41
5. Results ................................................................................................................................................. 43 5.1 Initial Data Analysis .................................................................................................................... 43 5.2 Optimized Routes and Corresponding Distances ........................................................................ 45 5.3 Margin and Cost Analysis based on Demand ............................................................................. 47 5.4 Scenario-based Sensitivity Analysis ........................................................................................... 49
6. Discussion ........................................................................................................................................... 63 6.1 Sensitivity Parameter-Based Analysis of the Results ................................................................. 63 6.2 Stakeholder Incentive Analysis ................................................................................................... 65 6.3 Recommendation ........................................................................................................................ 69 6.4 Contribution ................................................................................................................................ 70
7. Conclusion .......................................................................................................................................... 71 References ................................................................................................................................................... 73 Appendix ..................................................................................................................................................... 78
A. Amount of Plastic Generated by County .................................................................................... 78 B. County ID Mapping .................................................................................................................... 81 C. MRF ID Mapping ........................................................................................................................ 82 D. Amazon Warehouse ID Mapping ............................................................................................... 83 E. Cost, Price and Margin Calculation ............................................................................................ 84 F. Distances Matrix ......................................................................................................................... 87
8
List of Figures
Figure 1. Distribution of primary plastic production in different industries ............................................. 11 Figure 2. Spread of plastic waste production in different industries .......................................................... 12 Figure 3. Probability distribution of product lifetime across industries ..................................................... 13 Figure 4. Classic reverse logistics flow adapted from ............................................................................... 25 Figure 5. Volume of plastic sold by CPG companies in all of US by plastic classes ................................ 33 Figure 6. Per capita income for New England states relative to per capita income in the US ................... 34 Figure 7. Population ratio of New England states relative to US population ............................................. 34 Figure 8. Total plastics sold through CPG products in New England states .............................................. 34 Figure 9 Plastic sold by plastic classes in New England states .................................................................. 34 Figure 10. Lat-Long plot of County centroids, Amazon Warehouses and MRFs in the New England ..... 35 Figure 11. Flow of the post-consumer plastic based on model developed in this research ........................ 39
9
List of Tables
Table 5.1.1 Overall weight of plastics by annual sales in CPG Industry ............................................. 43 Table 5.1.2 Per capita income by New England states ......................................................................... 44 Table 5.1.3 Population ratio of New England states ............................................................................ 44 Table 5.1.4 Plastic sold by plastic type by county in New England (in Metric Tons) …..................... 45 Table 5.2.1 Distance Matrix …………................................................................................................ 46 Table 5.3.1 Aggregated cost and price calculation for all the counties by plastic classes ................... 48 Table 5.4.0 Parameters for sensitivity-analysis.................................................................................... 49 Table 5.4.1 Base case scenario for sensitivity analysis......................................................................... 50 Table 5.4.2 Lower transportation cost scenario for sensitivity analysis................................................ 51
Table 5.4.3 Larger service area within a county ................................................................................... 52 Table 5.4.4 Partnering to share logistics cost ....................................................................................... 53 Table 5.4.5 Impact of capacity of vehicle ............................................................................................ 54 Table 5.4.6.1 Impact of percentage of the vehicle capacity used in Type 1 vehicle ............................... 55 Table 5.4.6.2 Impact of percentage of the vehicle capacity used in Type 1 vehicle ............................... 56 Table 5.4.7.1 Impact of percentage of the vehicle capacity used in Type 2 vehicle ............................... 57 Table 5.4.7.2 Impact of percentage of the vehicle capacity used in Type 2 vehicle ............................... 58 Table 5.4.8.1 Emissions in Type 1 vehicle .............................................................................................. 60 Table 5.4.8.2 Emissions in Type 2 vehicle .............................................................................................. 60 Table 5.4.9 Impact of customer incentives .......................................................................................... 62 Appendix A Amount of Plastic Generated by County ........................................................................... 78 Appendix B County ID Mapping .......................................................................................................... 81 Appendix C MRF ID Mapping ............................................................................................................. 82 Appendix D Amazon Warehouse ID Mapping ...................................................................................... 83 Appendix E Cost, Price, and Margin Calculation ................................................................................. 84 Appendix F Distance Matrix ................................................................................................................ 87
10
1. Introduction
Plastic waste is one of the primary global challenges facing humanity and our environment
in the 21st century, creating intense inspection from consumers and industry into the life cycle of
non-biodegradable plastic (Verma, Vinoda, Papireddy, & Gowda, 2016), (Narancic & O’Connor,
2019), (Chow, So, Cheung, & Yeung, 2017). The mismanagement of plastic waste is polluting the
oceans, and this proliferation, if not checked, will add to the massive waste problem currently
threatening the world (Jambeck et al., 2015), (Verma et al., 2016); (Tammemagi, 1999). In 2017,
35.3 million tons of plastic was generated in the US, out of which 2.9 million tons were recycled,
5.6 million tons were incinerated, and 26.8 million tons, a staggering 75.8%, were landfilled, (US
EPA, n.d.). The incineration of the plastic impacts air quality, which further threatens the
environment and poses a significant threat to human beings unless it is managed in a controlled
environment, as in some of the Nordic countries (Fråne, Stenmarck, Gíslason, Lyng, & Løkke,
2014) and the UK (Jeswani & Azapagic, 2016). To decipher the magnitude of plastics being
introduced into the environment and the oceans, we need to understand the lifecycle of plastics
through processes such as, production, distribution, and waste management. Because of plastics’
persistence in the environment, we must consider not only last year’s production of plastic, but
also all plastic production over time, and its infusion into the environment.
Plastics can be broken down into two categories: fiber and non-fiber plastics. The primary
polymers that make up non-fiber plastics are Polyethylene (PE) (36% of global plastic production),
Polypropylene (PP) (21%), and Polyvinyl Chloride (PVC) (12%), followed by, in smaller
proportions, Polyethylene Terephthalate (PET), Polyurethane (PUR), and Polystyrene (PS) (<10%
each). Approximately 70% of all of fiber plastic production can be attributed to Polyester, most of
which is PET. These seven groups together amount to 92% of all plastics produced. Approximately
11
42% of all non-fiber plastics have been used for packaging, which is predominantly composed of
PE, PP, and PET (Geyer, Jambeck, & Law, 2017). Packaging plastics accounts for 40% of all
plastic produced, which is a staggering number and it is continuing to grow (Narancic & O’Connor,
2019).
Plastic production information helps us to understand the generation of plastic waste.
Figure 1 shows the plastic used by different industries between 1950 and 2015. The packaging
industry used the highest share of plastics and showed the biggest growth in production over time.
Figure 1. Distribution of primary plastic production in different industries. Adapted from (Geyer et al., 2017)
Figure 2 takes this a step further, showing the amount of plastic waste that has been
generated by the same industries. The packaging industry dominates the plastic consumption
market and thus the waste generation.
0
50
100
150
200
250
300
350
400
1950 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015
Mil
lio
n U
S T
on
s
Plastic Production by Industry over Time
Packaging Building and ConstructionTransportation Electrical / ElectronicConsumer and Institutional Products OtherTextiles Industrial Machinery
12
Figure 2. Spread of plastic waste production in different industries. Adapted from (Geyer et al., 2017)
A lifetime can be attributed to plastic packaging like the lifetime assigned to any products
which are in use; Figure 3 shows distributions across industries in terms of product lifetimes.
Ranging from toothbrushes to soap bottles, the plastics used in packaging have a particularly short
lifetime, often less than one year due to the quick consumption period, coupled with the recurring
nature of these products. These quick consumption times can be contrasted with plastics used in
the construction, automotive or information technology industry, where the consumption period
or lifetime can be in the range of years or decades. This dynamic has led plastics produced for
packaging in consumer-packaged goods (CPG) to particularly contribute to the proliferation of
global plastic waste.
0
50
100
150
200
250
300
350
400
1950 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015
Plastic Waste Generation by Industy
Packaging Building and Construction
Transportation Electrical / Electronic
Consumer and Institutional Products Other
Textiles Industrial Machinery
13
Figure 3. Probability distribution of product lifetime across industries (Geyer et al., 2017)
Therefore, there is a need to recycle or reuse the plastics in packaging and reduce the
production of new plastics globally (Hopewell, Dvorak, & Kosior, 2009). In response to this crisis,
many companies have started to evaluate new strategies to reduce plastic packaging waste, such
as including more post-consumer plastic in their product packaging, for example, the Alliance to
End Plastic Waste formed to start formalizing a solution to this global problem, and a sum of US
$1.5 billion has been pledged by the members of this consortium towards fighting the plastic waste
problem (“Alliance To End Plastic Waste,” n.d.).
One way to do fight the plastic problem is to improve the take-back of waste packaging in
order to reuse it in new packaging. Current recycling systems are broken in the US and there are
no effective mechanisms to take back plastic (Katz, 2019). Since China’s ban on taking plastic
waste from the US, municipalities are facing an even larger problem as to how to get rid of the
plastic waste that is produced in the form of municipal solid waste (MSW). A detailed 2020 study
suggests that only a certain percentage of plastics is being recycled depending on the type of
plastic, namely PET, high density polyethylene (HDPE) and PP (only 53%). The US doesn’t have
adequate capability to recycle other types of plastics (John Hocevar, 2020).
14
Based on this literature review, we can say that most post-consumer plastics in packaging
are of types PET and PP, and that we need an efficient mechanism to take them back for recycling.
To effectively improve the take-back of post-consumer plastic packaging waste, there is a need to
understand and model a closed-loop supply chain.
This thesis considers one mechanism that could contribute to this vision: a reverse flow of
plastic packaging waste using existing e-commerce distribution channels. By building a model
based on industry data and other predictable and measurable parameters, we were able to test the
feasibility, efficiency, and cost-effectiveness of this system.
This thesis is structured into seven chapters, beginning with this introduction. In Chapter
2, we present the problem statement and objectives. Chapter 3 provides an extensive review of
literature relevant to the proposed problem setting and methodology. Chapter 4 explains the
methodology adopted in detail, including formulation of the network design model, and
understanding the systemic cost equation. In Chapter 5, we outline the results from initial data
analysis, the optimization model implementation and the cost analysis, and the scenario-based
sensitivity analysis based on the results. In Chapter 6, we discuss the results from the scenario-
based sensitivity analysis, a qualitative study of stakeholder initiatives, provide recommendations
and explain the contributions. Finally, in Chapter 7, we conclude this thesis, discussing the
assumptions and touching upon the road ahead.
15
2. Problem Setting and Objectives
The primary goal of this research is to design a model to facilitate an e-commerce-based
reverse logistics channel approach to formulate a take-back of post-consumer plastic and thereby
assess the feasibility of the same from economic, social and environmental points of view. When
we order something online (say an Amazon order), in general, we expect the order to be delivered
to our doorstep. In the door delivery process, the delivery van could, instead of leaving empty-
handed after dropping the order, pick up post-consumer plastic and place it in a segregated section
in the van, effectively initiating a reverse logistics process to a material recovery facility (MRF)
directly or intermediary storage. This process can be made possible by any third-party logistics
provider.
The objective is to first identify the different parameters in the system, such as, various
costs, the volume of post-consumer plastic, and the price of different types of virgin plastic. The
object is also to identify various actors of the system. We start by analyzing the volume, value and
geographic distribution of the plastic sold by the CPG company. In terms of problem setting, we
consider the US plastic sales data and focus primarily on distribution within the New England
region, in states: Connecticut, Maine, Massachusetts, Rhode Island, New Hampshire, and
Vermont. We study the costs in several tranches of operation. We perform this analysis using the
sales data of products by a major CPG corporation as a case study, augmented by geographic
locations and distance data of warehouses of prominent e-commerce providers and MRFs utilizing
Google Maps API.
Then the objective of this research is to develop a network design model to assess the flow
of the plastic take-back from a county to an MRF using a direct path or using a consolidation
16
network, utilizing a warehouse (or distribution center) as a consolidator. Based on this model, we
assess the overall cost that the company facilitating this process might incur.
The next objective is to propose a cost equation to assess the feasibility of the optimization
model from the economic and environmental points of view. We assess the economic feasibility
based on cost equation and determine the profit margin based on the analysis per county for the
New England region. Then we assess the feasibility from the environmental point of view, by
studying the CO2 emissions as a result of this process and the cost of externalities by estimating
the cost using standard carbon tax estimates.
Finally, our objective is to consider the stakeholder ecosystem required for this model to
work. We identify the relevant stakeholders in the system and how each of the stakeholders could
be incentivized both from economic and social responsibility points of view.
17
3. Literature Review
This chapter aims to provide background information and review the existing literature
surrounding this project. This review includes:
(1) Policy directives pertaining to recycled plastic usage: Take-back policies for waste and
hazardous materials like electronics differ around the world, and we examine the relevant policies
that are in place for items like plastics that could impact the implementation of an e-commerce
take back system. We look for examples where recycling is mandated by the governments,
attempting to draw parallels for plastic packaging. We assess how similar policies can be designed
for the plastic recycling regulations and how companies could implement those models.
(2) The intricacies of customer behavior towards the use of plastic and recycling of plastic
packaging in CPGs: We ascertain that the customer is indeed concerned about the plastic pollution.
We use this consumer concern to evaluate the likelihood for consumers to participate in the take-
back process and assess the need to incentivize the consumer to return the post-use plastic
packaging to the retailer or the manufacturer.
(3) Potential uses of post-consumer plastics: We assess the recycling potential of plastic
packaging by categorizing the various types of plastics based on their potential recyclability. We
understand the potential uses to identify economic opportunities through the reuse, recycle and
remanufacturing methods, we discuss in Section 3.3.
(4) Existing reverse logistics mechanisms for products in other industries: We understand
how the take-back process through reverse logistics works for products in other industries like the
textile and electronics industries. Studying the existing reverse logistics mechanisms used for
recycling in recyclable substances would enable us to draw similarities in processes.
18
(5) Uses of the e-commerce reverse logistics channels for take-back: We study the
feasibility that e-commerce can be used to take-back plastic. This study of the existing e-commerce
reverse logistics channels helps us understand, how the existing flow of post-consumer products
or consumer returns can be fused with the post-consumer plastic take-back. We, also, verify that
this system has not been tried thus far and this is identified as the gap in the existing literature.
(6) The different types of costs: We focus on the different types of costs involved in the
several mechanisms affecting the take-back flow. As a last step, we assess the cost of the operations
of the take-back and the purchase cost of virgin and recycled plastics and how this makes the whole
process economically feasible. This assessment of different costs helps us formulate the profit
margin of the facilitating entity that enables the process recommended in this paper.
3.1 Policies on Plastics
There are several directives in place for several hazardous products spanning different
industries. End-of-life electronic products can result in hazardous e-waste, and hence there are
numerous directives for take-back of the products by the manufacturers. For the purpose of this
thesis, we draw parallels from the policy directives around e-waste take-back and look for similar
policy directives or the potential for such directives for plastics in the United States.
3.1.1 EU Directives on Electronic Items Take-Back
The EU has strong laws for the take-back of the end-of-life post-consumer electronic item,
and it is the producer’s responsibility to arrange to collect the items. This is known as the Extended
Producer Responsibility (EPR), and the boundaries of the same have been debated. These laws
have forced the companies to primarily think about four different strategies: (1) forming a take-
back network; (2) rethinking product design; (3) setting up a closed-loop supply chain; and (4)
adopting new technologies and business models. EU models generally stipulate what producers
19
must spend on the take-back of the products based on the market share of the producer. A good
model of this can be found among companies like Hewlett-Packard (HP Inc.), Sony Corporation,
Braun GmbH, and Electrolux AB. Apart from the cost incurred by the take-back of end-of-life
products from the customer, the costs for recycling are also borne by the producers based on the
market share of the producer. A further discussion stems from the context of implementation of
EPR. These laws do not encourage product innovation, which in turn reduces the need for
recycling. A study has suggested that producers pay for a share of the take-back based on the
percentage of their products which require take-back and recycling. This would encourage a long-
term focus on product innovation so that the need for the take-back is minimized (Atasu & Van
Wassenhove, 2011).
3.1.2 US Directives on Electronic Waste
Federal laws for take-back of electronic waste do not exist in the US; however, 22 out of
the 50 states have passed e-waste bills that mandate producer responsibility (Atasu & Van
Wassenhove, 2011). Some states in the US have implemented EPR-type regulations. In the US
few states that have mandated EPR for batteries, such as New Hampshire’s ban on disposal and
incineration of batteries (New Hampshire Code of Administrative Rules, 2017). EPR helps shift
the costs from the municipality to the producer, while at the same time enabling value extraction
if possible from the end of life.
3.1.3 EU Directive on Plastic
The EU effectively banned single-use plastic (Brussels 2019) in 2019 due to the ubiquitous
nature of the single-use plastic and its proliferation by short-term usage which causes pollution.
The EU member states have sparingly adopted this directive and are forming implementation and
enforcement strategies to combat the single-use plastic.
20
3.1.3.1 The EU directive is a step towards establishing a circular economy where
the design and production of plastics and plastic products fully respect reuse, repair and recycling
needs and more sustainable materials are developed and promoted. There are highly negative
impacts in terms of environmental, health and economic aspects from the use of certain plastics.
The environmental impact of toxins can cause health problems both in animals and humans
(Verma et al., 2016). Cancer incidents near MSW incinerators are also important factors to
consider while planning to mitigate plastic waste by burning (Elliott et al., 1996). Such negative
impacts require the setting up of a specific legal infrastructure to effectively mitigate these negative
impacts (General Secretariat of the European Parliament and of the Council, 2019).
3.1.3.2 The existence of policies that promote circular mechanisms to facilitate
take-back of toxic and hazardous products both directly and indirectly are in effect in the EU. The
policy triages effective non-toxic multi-use products, as opposed to single-use products, to reduce
waste generation and thereby mitigate pollution through waste. (General Secretariat of the
European Parliament and of the Council, 2019)
3.1.4 US Directives on Plastic Usage
No such federal laws exist so far in the US, but there is a strong inclination to ban single-
use plastic products like straws and plastic bags. For example, Boston has started the use of
reusable plastic bags and customers have been charged at least 5 cents for a reusable plastic bag
(Phillips, 2018). There are proposed federal policies like “Save Our Seas Act 2.0”, which aims at
improving response to marine plastic and also contribute at an international level to control the
advent of new plastic into the oceans. At the time of this writing, this act has passed through the
final stages of the Senate committee on Commerce, Science and Transportation (Whitehouse,
2019). At the time of writing this paper, another policy, “Break Free From Plastic Pollution Act”
21
has been placed in the Congress, and is yet to be approved. This policy establishes the following
guidelines: (1) minimum reuse, recycling, and composting percentages of products, and (2)
increasing the content of recycling material in new product manufacturing. This act would also
encourage producers to put easy to read labels and also if the product is reusable, recyclable or
compostable (Udall, 2020).
3.2 Consumer Response to Plastic Recycling and Take-Back Programs
The consumer is more willing to pay (WTP) towards plastic recycling costs than they are
for aluminum, glass and cardboard cartons. The customers’ WTP is assessed through the
embedded recycling cost in the product. However, consumers living in “bottle-return states” do
not express a higher WTP towards recycling costs. This is because of the expectation of bottle
return in the “bottle-return states” makes the inherent higher prices evident in the price for the
initial product purchase (Klaiman, Ortega, & Garnache, 2016). Environmentally friendly products
can have a positive impact on consumer choices, and green packaging drives consumer behavior
sufficiently to attract environmentally responsible customers to purchase greener products (Rokka
& Uusitalo, 2008). This customer behavior leads to the following: that consumers would think
positively about recycling of plastic and would participate in the take-back of the plastic packaging
of CPG products.
3.3 Recycling and the Potential Use of Collecting Post-Consumer Plastic
We discuss the potential use of post-consumer plastic and outlines the benefits of recycling
from the circular economy standpoint.
3.3.1 Drivers of Sustainable Plastic Solid Waste Recycling
At the household level the driver of recycling MSW is primarily to reduce the creation of
waste that doesn’t decompose (Tonglet, Phillips, & Bates, 2004). At a psychological level,
22
minimizing waste creation is more powerful to adhere to for the consumer than a local government
mandated requirement to recycle, and thus programs geared towards exciting monetary
opportunities to reduce waste pushes households to recycle more and also create less waste
(Tonglet et al., 2004). Consumers usually need to be educated to see the MSW as a resource with
an economic value attached to it, however in the US the benefits of recycling have long been
promoted.
3.3.2 Economic and Environmental Motivation for Fossil Fuels
As virgin plastic is typically created from fossil fuels, recycled plastic can reduce the
manufacture of virgin plastic, thus saving petroleum, natural gas, and other byproducts. Also,
environmental protection through reduction of plastic manufacturing triggers consumer sentiments
and awareness towards being sensible about plastic use and plastic recycling. Moreover, both
consumer and producer responsibility rules and regulations have also been identified as drivers of
solid waste management systems from the economic, social and environmental aspects (Mwanza
& Mbohwa, 2017). As an economic driver the take-back plastic can be recycled and reused in
remanufacturing processes, reducing raw material costs in the process. On the environmental side,
regulation on plastic waste collection involves large-scale social endeavor directed towards an
environmental cause, as societies come together to facilitate recycling and be an active participant
in the process. Similarly, as an environmental driver, the regulations protect the environment (and
society) from the toxins released by plastic waste when landfilled or incinerated.
3.3.3 Future Use of Post-Consumer Plastic
A theoretical study suggests that any product take-back can have multiple benefits for the
manufacturers, such as (1) a source of inexpensive components and materials; and (2) avoidance
of disposal and incineration costs to be incurred by the producer based on EPR policies discussed
23
in Section 3.1.2; and (3) a buy-back opportunity for manufacturers to sell new products, such as
polyester-based clothing material that is very popular for athletics and other sports. New products
could also entail substituting recycled plastic in products originally made from virgin plastics.
Thus, the study bolsters our assumption that there will be a monetary value associated with product
recovery by the producer. (Thierry, Salomon, van Nunen, & van Wassenhove, 1995)
3.4 Use of Reverse Logistics in Take-Back for Recycling
After the discussion on plastic take-back and its benefits in prior sections, we now study
where reverse logistics has been used for returns and take-back of products. We look at textile and
battery take-back as examples to draw parallels and similarities to our model of the plastic take-
back.
3.4.1 Similarities of Post-Consumer Plastic Take-Back for Recycling with Textile Take-Back
Processes that are like those in a proposal to use reverse logistics of textile (Bukhari,
Carrasco-Gallego, & Ponce-Cueto, 2018) can be understood, and expanded, for plastics. The way
each type of plastic is collected from the end consumer determines how complex the system might
be designed. Expanding upon a general consolidation-based network design, we can understand
how e-commerce (and other reverse logistics channels) can be used to take back the plastic to a
sorting location. Furthermore, the use of upcoming artificial intelligence (AI) based computer
vision technologies like AutoSort, which uses robotics to sort between visibly different substances
(Hahladakis & Iacovidou, 2018). This can be used to sort different types of plastics, for example,
this technology can be used to segregate bottles (PET) and caps (PP). This process further helps
the recycling processes, as the process to recycle PET is different from PP.
3.4.2 Similarities of Post-Consumer Plastic Take-Back for Recycling with Battery Take-Back
24
Process similarities of post-consumer plastic take-back for recycling with battery take-back
is studied in this paper. This research uses a mixed-integer linear programming (MILP) based
network design model from the consumer location to a sorting center and then to a recycling plant
can be assessed as one of the potential mechanisms for post-consumer plastic take-back. This can
be understood as a mechanism that drives e-commerce-based reverse logistics, where a return is
picked up from an end-consumer, consolidated at a warehouse or a distribution center and then
sent to the manufacturer (Ponce-Cueto & González-Manteca, 2012). We can leverage a similar
model while designing the take-back of post-consumer plastic for recycling.
3.5 Use of E-Commerce in the Take-Back Process
In this section, we study e-commerce, primarily from the reverse logistics standpoint. We
understand customer returns and the process of e-commerce take-back to facilitate returns. There
are several models, such as a consumer-based return aggregator (e.g., Amazon Hub Locker), direct
pickup (e.g., UPS pick-up) from consumer locations, and consumers sending the product back
through common logistics providers (e.g., FedEx, UPS, US Mail and others).
3.5.1 E-Commerce Returns
E-commerce reverse logistics channels has been used to facilitate the customer returns
process primarily. However, it has also been used to support the following: (1) competitive
advantage – efficient handling of returns of the products in the e-retail space can generate large
cost savings; (2) product reuse – effective use of reverse logistics for the return of the product
facilitates reuse. This enables value extraction from the product, by direct reuse or by generating
value by disintegrating the parts when the returned product is put through the remanufacturing
process; and, (3) environmental impact – adhering to the EPR in the EU to reduce the volume of
waste (Kokkinaki, Dekker, de Koster, Pappis, & Verbeke, 2002).
25
3.5.2 Process of the E-Commerce Take-Back
To understand the e-commerce-based take-back process it is important to understand the
material flow of the product in such a system, as Figure 4 shows.
Figure 4. Classic reverse logistics flow adapted from (Kokkinaki, Dekker, De Koster, Pappis, & Verbeke, 2002)
The supply chain of the product flow is important to understand to understand the reverse
flow of the products. The forward flow starts after initial product manufacturing. The product
flows from the factory to warehouses or distribution centers, where it is stored to be further shipped
to stores or directly to customers (in case of e-commerce). Finally, the product reaches the
customer through retail or e-commerce channels. The reverse logistics process starts when the
customer initiates a return on a used or unused product. The product is either picked up from the
customer location or the customer drops the product off at a drop-off location. The product is then
carried to a consolidation center, usually a warehouse or a distribution center. The product
undergoes inspection through a sorting and selection process. Then, after sorting and selection, it
is determined which of the returned products will be reused, recycled or remanufactured, or which
products will be disposed of. Based on this decision the products move to redistribution after the
26
completion of the aforementioned processes (Kokkinaki et al., 2002) (Govindan, Palaniappan,
Zhu, & Kannan, 2012).
3.6 Aspects of Cost in the Take-Back Mechanisms
Finally, in this section of the literature review, we study the costs of operations. The costs
of operations signify the various component costs to enable a take-back process using reverse
logistics channels. We also understand the various other costs in terms of recycling processes, and
operational costs in the MRFs.
3.6.1 Cost of Reverse Logistics Modes to Decide Optimal Take-Back Channels
To utilize the reverse logistics channels to facilitate take-back it is important to understand
the cost in each of the reverse collection channels. Every step of the supply chain incurs cost.
Focusing primarily on the reverse logistics supply chain the costs can be summarized as pickup
cost, transportation cost (primary leg, middle-mile and last-mile), sorting and handling costs at the
warehouse, storage cost, and other miscellaneous costs such as IT, human resources, etc. The
optimal reverse logistics route is typically selected based on the minimum total cost incurred in
that route as compared with the total cost incurred in all other routes. Studies show that, in the case
of products manufactured by Apple Inc., HP Inc., and The Eastman Kodak Company, the choice
of the optimal reverse logistics channel strongly depends on the cost of the channel, type of the
product, and the volume of units sold. Because of the economy of scale, the take-back through the
retailer is more cost-effective; however, in the case of fragmented dissemination of products and
brands, a manufacturer take-back is more cost-effective (Atasu, Toktay, & Van Wassenhove,
2013). Even though we can pull similarities from this outcome we cannot comment on whether
27
the same pattern will be applicable in case of the plastic product take-back; more research is needed
to better understand this dynamic for plastic (Klausner & Hendrickson, 2000).
3.6.2 Cost of Recycling
There has been limited research regarding the cost of recycling processes in the US. The
cost of recycling is dependent on several variables, such as collection techniques, frequency of
collection, equipment used, and the type of material that is collected for recycling (Hegberg,
Hallenbeck, & Brenniman, 1993). This study also showed approximate costs of collection of
different types of plastic per household per year and the breakdown of the recycling rates.
However, this research is dated, from 1993, and thus, the cost figures mentioned in the study would
not be relevant in the current scenario and the cost of recycling would be needed to be considered
from recycling plants’ current price quotations.
3.7 Conclusion of Literature Review
In the literature review, we found that there are no federal policies for plastics in the US,
however, the general household is more attuned to this global problem and shows more empathy
towards plastic recycling and willing to pay more for plastic packaging in lieu of recycling costs.
We studied the potential of the plastic take-back and we discovered several opportunities that post-
consumer plastic can uncover. We found that take-back policies for different products have worked
out well in the past using reverse logistics channels. We also, found that studies have been
conducted to understand several implications of product take-back, some stipulated by laws, others
to generate value from the post-consumer product. In the case of plastic, post-consumer plastic can
be a viable option for value generation for companies that facilitate the take-back process.
28
The literature review presented in this section shows that the literature and research on
using e-commerce models for plastic take-back is scarce. The gap in the literature is that the
assessment of using e-commerce channels for the take-back of post-consumer plastics generated
from CPG products has not been done. This gap has been identified in the literature review done
in this research. This research, thus, aims to shed light on the feasibility of such a model using the
e-commerce based reverse logistics channels.
29
4. Methodology
In the literature review we identified the gap in the literature regarding the use of e-
commerce channels to facilitate the take-back of plastic from consumer locations back to MRFs.
We also studied how a reverse logistics network has been used to facilitate the take-back of similar
waste generating products.
In this section we define the methodology and the steps we took in conducting this research.
This section can be broken down into seven actions: (1) Data collection; (2) Data preparation; (3)
Initial data analysis (4) Problem formulation using a network design approach; (5) Cost analysis
(6) Scenario based sensitivity analysis, and (7) Recommendations.
4.1 Data Collection
In this step we collected data from several sources regarding the following 10 topics:
(1) CPG product sales information – This gave us the total plastic waste generation by the CPG
company in a year through the number of products sold via and the weight of each plastic type in
tons;
(2) CPG product market share information – This product-specific market share information helps
us to estimate the overall US market for that product, making it useful for calculating the total
weight of plastic generated by the overall CPG industry by plastic type in tons;
Data
Collection
Data
Cleaning &
Prep-
aration
Initial Data
Analysis
Problem
Form-
ulation &
Network
Design
Cost
Analysis
Scenario
based
Feasibility
Analysis
Recom-
mend
Strategy
30
(3) Census information regarding population ratio per county – This data point allows us to
estimate the population ratio of every county on the US; and based on this information it will be
easy to estimate the consumption per county. [Data source: Census.gov];
(4) Census information regarding per-capita income per county – This data point allows us to skew
the plastic consumption information further to understand the actual consumption per county more
closely. This data point is applied over the population ratio metric to come up with the final plastic
waste numbers for every county. [Data source: Census.gov];
(5) County centroid points – County centroid points are latitude-longitude (Lat-Long) values that
generate a central point in the county based on data provided by Google Maps API. This data is
used to estimate the transportation miles for the local distances within the county. Data source:
Google Maps;
(6) MRF locations across the US – This data gives us the Lat-Long values for all the MRFs which
were further used to calculate the linehaul distances. [Data source: (“Residential MRFs - The
Recycling Partnership”)];
(7) Amazon warehouse locations across the US – This data point also, helps us to calculate linehaul
distances between county centroids and MRFs. We have used Amazon as a case study here due to
the number of warehouses in the US and because Amazon is among the most prominent e-
commerce actors in the US. [Data source: (“Locations of Amazon Fulfillment Centers in USA -
Forest Shipping,” n.d.) ];
(8) Operational cost information – Operational cost information takes into consideration different
costs that incur in different tranches of the operations. This cost information can be broken down
into several other data points such as (a) Cost of Transportation (US $/mile), (b) Cost of Storage
31
(US $/lbs.), (c) Cost per Stop (US $), (d) Cost of Recycling (US $/ton), (e) Cost of Emissions (US
$/ton-CO2). All these costs are relevant to understand the different scenarios and overall benefit of
using the take-back process for post-consumer plastic;
(9) Vehicle information – This data point specifically points towards understanding the various
types of vehicles, for example small e-commerce delivery vans with a capacity of 3,500 lbs. and
long-haul trucks with a capacity of 720,000 lbs.; and,
(10) Emission information – In this data point we estimate the total grams of greenhouse gases,
using the standard unit of CO2-equivalents (CO2e), generated by different types of vehicles using
the accounting methodology and average industry data for US specified within Global Logistics
Emissions Council Framework. In our research we primarily focus on small vans (vehicle Type 1)
and large trucks (vehicle Type 2) and consider both the weight of plastics transported and the
distance traveled.
This step enables us to move to the Data Cleaning and Preparation phase, which will make
the data ready for analysis.
4.2 Data Cleaning and Preparation
After data collection, we prepared the data by performing the following steps:
(1) Data cleaning – We eliminated missing data from the collected datasets, nameless from the
sales information and census information;
(2) Unit normalization – We performed unit normalization across our entire datasets to curtail
disparities between data collected through different channels. For example, we changed all the
weight values to US tons and smaller units to pounds similarly, we changed all the distances to
miles. Furthermore, we normalized all the dependent variables that depend on the weight and
32
distance values. For example, we changed the Cost of storage from $ per kg to $ per pound and
the Cost of Transportation from $ per km to $ per mile;
(3) Calculating overall US sales of CPG products – As discussed in Section 4.1 (Data Collection)
we calculated the overall US sales of the products sold by the CPG company by dividing the CPG
Company sales with their market share. This number gives us the total weight of plastic in the
products sold by the entire CPG industry. This also enables us to cluster the weight as derived
from sales based on plastic type to get the tonnage generated by specific plastic classes;
(4) Normalization on Sales Data – As discussed in the Section 4.1 we performed a normalization
operation on the overall US CPG sales data by plastic class by multiplying the tonnage with
population ratio and the income skew. The income skew was calculated by taking a weighted
average of county-specific per-capita income over the per-capita income of all of US;
(5) Preparing Distance Data – From the Lat-Long values collected for county centroids, Amazon
warehouses, and MRFs as discussed in Section 4.1 we calculated the actual distances by using the
Distance Matrix API provided in the Google Maps API suite. We wrote software code to invoke
the API iteratively to get a Cartesian product of distances against all the Lat-Long values, as
discussed in Section 4.1;
(6) Preparing Emissions Data – we parameterized the emissions data based on the values from the
data collected for two vehicle types mentioned in Section 4.1.
Upon completion of the data cleaning and preparation we could move to the initial data
analysis phase.
33
4.3 Initial Data Analysis
After the completion of data cleaning and preparation we performed an initial data analysis,
including data sensing, to understand the different clusters in which the data is spread out. For
example, we found the spread of the plastics collected over different plastic classes as shown in
Figure 5.
Figure 5. Volume of plastic sold by CPG companies in all of US by plastic classes
Similarly, we tried to understand the relative per-capita income and population ratio for all
the states in the New England region, as shown in Figures 6 and 7.
Data Sensing - Volume by Plastic Class
HD-POLYETHYLENES LD-POLYETHYLENES
POLYESTERS POLYPROPYLENES
34
Figure 6. Per capita income for New England states relative to per capita income in the US
Figure 7. Population ratio of New England states relative to US population
We understand how the plastic weight is spread across different states in the New England
area as shown in Figure 8 and investigate the plastic data through a further breakdown analysis by
plastic classes as shown in Figure 9.
Figure 8. Total plastics sold through CPG products in New England states
Figure 9 Plastic sold by plastic classes in New England states
0 0.5 1 1.5
Maine
Vermont
Rhode Island
New Hampshire
Massachusetts
Connecticut
Relative per-capita income
0 1 2 3
Vermont
Rhode Island
Maine
New Hampshire
Connecticut
Massachusetts
Population Density
0 10000 20000
Vermont
Rhode Island
Maine
New Hampshire
Connecticut
Massachusetts
Total plastics sold by tons
0 5000 10000 15000
Vermont
Rhode Island
Maine
New Hampshire
Connecticut
Massachusetts
Plastic classes by weight sold
LDPE HDPE PP PET
35
We also plot all the physical locations of county centroids, Amazon warehouses and MRFs
on the US map based on their Lat-Long values as discussed in Sections 4.1.5, 4.16 and 4.1.7.
After the initial data we formulate our model using the network design approach, as
described in Section 4.4.
4.4 Problem Formulation Using A Network Design Approach
After completing the initial data analysis, we used what we learned to formulate our model
using a network design approach.
Figure 10. Lat-Long plot of County centroids, Amazon Warehouses and MRFs in the New England area
36
In this formulation we designed a network design optimization model using a mixed-
integer linear programming approach to minimize the logistics cost. The logistics cost is a
combination of the cost of transportation (Section 4.1.8a) and the cost per stop that the third-party
provider incurs (Section 4.1.8c) while operating this model. This cost is largely defined by the
transportation costs, which includes local delivery rounds, where the pickup vehicle makes a
number of stops, and line-haul transport, where consolidated packages make a longer trip from the
county centroids to the warehouse (Leg 1), directly from county centroids to the MRFs (direct)
and warehouse to the MRF (leg 2).
This model considers cij, the logistics cost calculated based on Clogistics in (7), which feeds
into the optimization model formulation in (1). xij demonstrates the quantity of plastic collected,
and z is the binary parameter which determines if the model should choose a direct path from the
source (County) to the destination (MRFs) or it should choose a consolidation route through a
warehouse of a third-party logistics provider or an e-commerce provider. The constraints are
delineated from (2) through (6). The constraint in (2) is a binary parameter that decides whether a
direct route is chosen, or a consolidation route is chosen. Constraint (3) describes the origin volume
constraint, which can be explained as the volume that is considered from an origin point (county)
cannot more than the post-consumer plastic generated at that point. Furthermore, (4) explains the
capacity constraints on the intermediary point and the termination points of the route. Equation
(5), explains the transshipment constraint which entails the number of pounds coming into a
transshipment facility, leaves the facility in its entirety. Constraint (6) is a non-negativity constraint
on the amount of material in flow from the source to the destination.
37
"#$(&) =))*+!",!" +
#
"$%
&
!$%
))(1 − *)+!',!' +))(1 − *)+'",'"
#
"$%
(
'$%
(
'$%
&
!$%
… (1)
Subject to:
* = 20,+!'++'" <+!"1, +!'++'" ≥ +!"
, ∀8 ∈ $), : ∈ $* , ; ∈ $+ … (2)
),!"-).((/0#,∀!∈4%!
… (3)
),!"-567(78!90&,∀"∈{4',4(},<=5=%"
… (4)
),!' =!
),'""
, ∀8 ∈ $), : ∈ $* , ; ∈ $+ … (5)
,!" ≥ 0 … (6)
A/>?!@9!8@ = A@ BC +D
EF#7A+12G +AB B2 B
DEF#7A
+12G H/!&CD7./ +
C:+)E√J
G… (7)
where,J = HPCQ8RS = FC#7&B
GHC7=
F
IH)… (8)
"=capacityofthevehicle
This model presents the opportunity to several sensitivity parameters to evaluate different
scenarios in terms of operations cost and overall profit of each scenario. These sensitivity scenarios
help us to dynamically assess several factors. For example, a region can be categorized as rural
and urban habitation based on the population ratio in terms of number of households per square
38
mile. Similarly, other parameters aid the understanding of different scenarios based on geographic,
social and economic likelihoods.
Sensitivity Parameters
n = number of households (collection points)
Qmax = Capacity of the transportation vehicle
r = radius of the area considered
D = Demand (based on company sales data, population ratio, and per-capita income of the
county)
" = % capacity of the vehicle used
This formulation enables us to perform the optimization to understand, based on the
location of the warehouses and MRFs, which leg of transport is the most cost effective. After the
formulation of this model we to perform cost analysis for the company enabling the process of
take-back.
4.5 Cost Analysis
After devising the model using a network design approach and coming up with the
transportation routes, we now calculate the profit margin for the company facilitating this take-
back process. In this we take the perspective of the CPG company and assume that the CPG
company is facilitating this take-back process. However, this analysis will hold good for any entity
that facilitates this take-back process, such as a logistics service provider or recycling company
who will plan to sell the recycled plastic to plastic manufacturers.
To perform this cost analysis, we consider the following flow of actual post-consumer
plastic from the consumer to the CPG company through various processes.
39
To perform the cost analysis, we have come up with a generic equation. The equation below
represents a mathematical formulation, which takes into consideration the purchase price of virgin
and recycled plastic. The formulation suggests a parameter a that varies between 0 and 1 and
determines the component structure of the products of the CPG company. It estimates the price
that the CPG company will not have to pay if they undertake this process of facilitating the take-
back of plastics and thereby facilitating recycling, and then collects and uses the recycled plastic
pellets to manufacture future plastic packaging.
In this equation, we also consider a total cost, which is composed of the following costs,
as covered in Sections 4.1.8 and 4.4: (1) Recycling cost, (2) Logistics Cost, (3) Cost of Storage,
(4) Cost of Sorting (usually included in the recycling cost), and (5) Parameterized cost of
incentives.
!"!"#$"% + (1 − !)"#&'(')&* − ()*+,.)/* = 1+2345 −67(.&%!)
Figure 11. Flow of the post-consumer plastic based on model developed in this research
40
where,
0<U<1,
()*+,.)/* = .#&' +.)+$",-"', +.,-+#.$& +.,+#-"%$ + ."%'&%-"!&,
where,
Pvirgin= Purchase Price of virgin plastic
Cenv= Estimated environmental cost
Precycled= Price of Recycled Plastic
Crec= Recycling costClogistics= Total Logistics CostCsorting= Sorting Cost
Cincentives= Incentives Costs
Cstorage= Storage Cost
After conducting the cost analysis and applying the formula to the modeled data, we then
conduct a scenario-based sensitivity analysis.
4.6 Scenario-Based Sensitivity Analysis
After completing the cost analysis, we perform the sensitivity analysis based on different
sensitivity parameters, as mentioned in Section 4.4. In this sensitivity analysis we change the
different parameters to understand the impact on the profit margin of the entity facilitating the
take-back process. For this research we do the sensitivity analysis from the perspective of the CPG
company that is facilitating the process.
41
To conduct the sensitivity analysis, we use the results from the cost analysis for all the
counties in the New England states, and plot them in four different graphs showing each of the
following relationships: (1) margin across all the counties; (2) different types of costs across all
the counties; (3) specifically logistics cost across the counties; and (4) emission cost vs. margin
across all the counties.
Based on this analysis we can understand which scenarios work well from an economic
perspective and how the choice of distance and vehicle affect the greenhouse gases emitted from
the transportation required by this process. The effect of emissions is further analyzed based on
the cost to the company using a carbon price ($ per ton-CO2). The results from the sensitivity
analysis is detailed in Section 5.4.
In our analysis, however, we do not subtract the emissions cost from the margin. We show
it separately as this can be further acted upon using various other measures, such as carbon offsets
and the cost of investment in an electric fleet.
After conducting the sensitivity analysis, we are poised to make recommendations to the
company facilitating this take-back process.
4.7 Recommendations
After performing the sensitivity analysis, we make recommendations to the entity
sponsoring this take-back process based on what parameters to choose to maximize economic
benefit while minimizing emissions and ensuring greater plastic collection. The plastic collection,
however, is dependent on customer responsibility, which can be further assessed using “pay or
punish” model. A deeper understanding of incentives can help in assessing if there is a relationship
between collection percentage and stakeholder incentives.
42
We used this methodology (Section 4) to conduct our studies and calculations and reached
the results discussed in the next section.
43
5. Results
After discussing the methodology for this research, we now discuss the results that were
obtained by conducting the analysis on the data collected. These results and sensitivity analysis
present the outcome of running the optimization-based network design model and the cost analysis
defined in the chapter 4, Methodology.
These results are broken down into initial data clustering based on product types and
contents, and geographic distribution of locations in terms of counties, Amazon warehouses, and
material recovery facilities (MRFs). The results also demonstrate optimized route distances, the
cost structures and the profit margin as described in the Methodology section.
5.1 Initial Data Analysis
Upon executing the methodology as mentioned in Chapter 4, we find several interesting
insights from the initial data analysis as described in Section 4.3. We first found the weights of
different types of plastics as described in Table 5.1.1.
Table 5.1.1 Overall weight of plastics by annual sales in CPG Industry
Type of Plastic Metric Tons HD-POLYETHYLENES 9820.39 LD-POLYETHYLENES 204.81 POLYESTERS 220504.00 POLYPROPYLENES 304792.00
We see that polyesters (the majority of which is PET) and polypropylenes dominate the
post-consumer plastic space.
44
Next, we understand the relative per capita income among the New England states and see
that Connecticut has the highest relative per-capita income, followed by Massachusetts, and then
by New Hampshire, Rhode Island, Vermont and Maine. Table 5.1.2 demonstrates this.
Table 5.1.2. Per capita income by New England states
State Per-capita Income Per-capita Income density
Maine $ 48,905.00 0.898 Vermont $ 54,173.00 0.995 Rhode Island $ 54,850.00 1.007 New Hampshire $ 61,294.00 1.126 Massachusetts $ 71,683.00 1.317 Connecticut $ 76,456.00 1.404
Similarly, we see that the population ratio of Massachusetts is the highest followed by
Connecticut and then by New Hampshire, Maine, Rhode Island and Vermont. This is demonstrated
in Table 5.1.3
Table 5.1.3. Population ratio of New England states
State Population ratio Vermont 0.19 Rhode Island 0.32 Maine 0.41 New Hampshire 0.41 Connecticut 1.09 Massachusetts 2.11
After finding the population ratio and the relative per-capita income density, we can then
calculate the normalizing parameter which, when multiplied with the sales values, can give a near
estimate of the weight of products sold in specific counties. A snapshot of the data is provided in
Table 5.1.4, where we show the state, the counties, and the normalized weights of the plastics of
different types. The full table can be seen in Appendix A.
45
Table 5.1.4 Detailed distribution of plastic sold by plastic type by county in New England (in Metric Tons)
States and Counties Total Plastic PET PP HDPE LDPE
Connecticut 10519.91 4333.25 5989.65 192.99 4.025 Fairfield County 4384.79 1806.13 2496.54 80.44 1.68 Hartford County 2229.71 918.44 1269.51 40.90 0.85 Litchfield County 453.05 186.62 257.95 8.31 0.17 Middlesex County 430.22 177.21 244.95 7.89 0.16 New Haven County 1871.09 770.72 1065.33 34.32 0.72 New London County 608.89 250.81 346.68 11.17 0.23 Tolland County 329.07 135.55 187.36 6.04 0.13 Windham County 213.02 87.75 121.29 3.91 0.08 Maine 2520.87 1038.37 1435.29 46.25 0.96 Androscoggin County 172.21 70.93 98.05 3.16 0.066 Aroostook County 106.58 43.90 60.68 1.96 0.04 Cumberland County 706.90 291.18 402.48 12.97 0.27 Franklin County 44.39 18.28 25.27 0.81 0.02 Hancock County 108.18 44.56 61.60 1.98 0.04
After data preparation we run the network optimization model, and the results of which are
mentioned in the next section.
5.2 Optimized Routes and Corresponding Distances
After the initial data analysis and data preparation we ran the optimization to get the routes
from every county to the MRF. This process was executed in detail as described in Section 4.4.
To run the model, we assumed that the facilities in the model, e.g., the Amazon warehouses
and the MRFs, have infinite capacity. Thus, the facilities selected by the model to form a route
were primarily chosen based on the minimum distance as the transportation cost and the cost to
stop (as described in Section 4.1.8) were negligible for the local distances within the service radius
in the county.
Furthermore, we saw that all the distances selected are direct distances (shortest feasible
distance) due to the distance minimization (as described in Section 4.1.8). To normalize this, we
46
break down the consolidation distance in Leg 1 and Leg 2 and capture the consolidation route if
the Leg 2 distance is less than the direct distance. This logic signifies that if the CPG company
were to employ a 3PL provider, it will only do so if the Leg 2 distance is shorter than the direct
distance. In this case the transportation cost incurred in the Leg 1 distance is an additional cost the
CPG company is willing to incur due to the benefits of consolidation, which results in overall cost
reduction.
In Table 5.2.1 we show the selected distances for the counties, some of which are direct
and the remaining are through a consolidation network. The Table 5.2.1 shows county IDs, MRF
IDs, and Amazon Warehouse IDs, which are identifiers to represent a county, an MRF and an
Amazon warehouse, respectively, and this has been utilized for easy of multi-functional data
analysis. The full description of these Ids, can be found in Appendices B, C and D. Table 5.2.1
shows a snapshot of the data. The entire table can be seen in Appendix F.
Table 5.2.1 Distances in Miles between County and MRF (on left) and Distances between County and MRF Through Amazon Warehouse (right)
CTY_ID
MRF_ID Miles
Final for
Cost
Total Distance
(including Leg 1)
CTY_ID
AMZ_ID MRF_ID Miles CTY_CT_1 MRF_CT_4 12.14 12.14 12.14 CTY_CT_1 AMZ_4 MRF_CT_3 59.21
CTY_CT_2 MRF_CT_8 14.63 13.19 25.04 CTY_CT_2 AMZ_5 MRF_CT_7 25.04
CTY_CT_3 MRF_CT_4 34.15 13.91 57.59 CTY_CT_3 AMZ_3 MRF_CT_5 57.59
CTY_CT_4 MRF_CT_5 24.15 13.91 35.21 CTY_CT_4 AMZ_3 MRF_CT_5 35.21
CTY_CT_5 MRF_CT_3 22.08 21.26 30.95 CTY_CT_5 AMZ_4 MRF_CT_3 30.95
CTY_CT_6 MRF_CT_6 20.02 13.19 64.56 CTY_CT_6 AMZ_5 MRF_CT_7 64.56
CTY_CT_7 MRF_CT_6 12.45 12.45 12.45 CTY_CT_7 AMZ_5 MRF_CT_7 37.09
CTY_CT_8 MRF_CT_6 16.26 13.19 58.65 CTY_CT_8 AMZ_5 MRF_CT_7 58.65
CTY_MA_1 MRF_MA_1 33.35 27.58 90.65 CTY_MA_1 AMZ_2 MRF_MA_2 90.65
CTY_MA_10 MRF_MA_9 15.41 1.63 32.14 CTY_MA_10 AMZ_6 MRF_MA_4 32.14
47
This table clearly shows the route choices made and the final distances to be used in the
total systemic cost calculation, results of which we discuss in Section 5.3.
5.3 Margin and Cost Analysis based on Demand
After finding the optimal distances from every county to the closest MRF, we further
calculated the different components of the cost: the transportation cost, the stop cost, the overall
logistics cost, recycling cost, and the incentive cost (with the value of incentives as zero dollars to
begin with). We also calculate the price based on the weight of the post-consumer plastic for every
plastic class. This is important to understand what the CPG company would have spent to
manufacture the product packaging using virgin plastic. This finally brings us to calculating the
profit margin, which is calculated by subtracting the different costs from the price of the virgin
plastic.
48
Table 5.3.1 shows a snapshot of the whole calculation that was performed. The full table can be found in Appendix E.
This analysis shows the different costs and the margins for each individual county based on the calculated price of virgin plastic
and the summation of all the costs included here. Furthermore, we also estimate the emissions based on the number of trips and weight
carried per trip and through different vehicle type. We then calculate the emissions cost based on global average price of mandated
carbon taxes (The World Bank, 2020). We now perform a sensitivity analysis to understand the aggregated behavior of this system.
49
5.4 Scenario-based Sensitivity Analysis
As described in the methodology, we show the results from the sensitivity analysis
performed by varying the various sensitivity parameters in Table 5.4.0:
Table 5.4.0: Parameters for Sensitivity Analysis Transport Cost ($ / mile) Number of Households (units) Storage cost ($) Capacity of the vehicle (lbs.) Cost of recycling ($ / ton) Percentage of capacity used (%) Incentive Cost ($ / Household / Month) Distance negotiated (Yes / No) Radius of coverage (miles) Type of vehicle (Type 1 or Type 2)
In all the below scenarios we also consider that the collection is 100 percent which means
that the amount of plastic that is sold (in tons) is collected from the consumer after use through
this take-back process.
While conducting the sensitivity analysis we first consider a base case, as described in
Table 5.4.1 and other different cases as described in Tables 5.4.2, 5.4.3, 5.4.4, 5.4.5, 5.4.6.1,
5.4.6.2, 5.4.7.1, 5.4.7.2, 5.4.8.1, and 5.4.9. The base case scenario is decided based on the generic
use cases and the most commonly used scenarios. Table 5.4.1 and others as mentioned before is
composed of parameters which are described in Table 5.4.0. The afore mentioned Tables also
consists of Results in terms of Total Cost incurred, and Total Margin, which suggests a positive or
a negative margin, and Emissions Cost, which is incurred based on the vehicle choices while
conducting the sensitivity analysis. Furthermore, the plots in the Table 5.4.1 represent: (1) Margin
in US dollars over all the counties in New England; (2) Cost in US dollars over all the counties in
New England; (3) Emissions Cost over all the counties in New England; and (4) Logistics Cost,
which is a specific component of the total cost. These plots show vividly how the different choice
of parameters changes the nature of the plots.
50
5.4.1 Base Case Scenario
Table 5.4.1 shows the base case scenario results for sensitivity analysis.
Table 5.4.1: Base case scenario for sensitivity analysis Transport Cost ($/mile) - $2.38 /mile Number of Households (units) - 50 Storage cost ($) - $0.69 Capacity of the vehicle - 3,500 lbs. Cost of recycling ($/US ton) - $120 Percentage of capacity used - 20% Incentive Cost ($ / Household / Month) - $0 Distance negotiated - No Radius of coverage – 5 miles Type of vehicle (for emission cost) – Type 1
Results Total Cost $28,877,785.74
Total Margin $17,565,058.04 Total Emissions Cost $11,286,808.44
Plots
5.4.2 Impact of Transport Cost
In Table 5.4.2, we consider a lower transportation cost of $ 1.7 per mile, and the results are
as expected. The margin is higher, and both the overall cost and logistics cost are lower. This is an
51
important scenario, as we show that if the CPG company can negotiate the transportation cost with
the logistics provider, this venture becomes even more profitable.
Table 5.4.2: Lower transportation cost scenario for sensitivity analysis Parameters
Transport Cost ($/mile) - $1.70 /mile Number of Households (units) - 50 Storage cost ($) - $0.69 Capacity of the vehicle - 3,500 lbs. Cost of recycling ($/US ton) - $120 Percentage of capacity used - 20% Incentive Cost ($ / Household / Month) - $0 Distance negotiated - No Radius of coverage – 5 miles Type of vehicle (for emission cost) – Type 1
Results Total Cost $22,270,387.39
Total Margin $24,172,456.39 Total Emissions Cost $11,286,808.44
Plots
52
5.4.3 Impact of Service Radius and Number of Households Per Trip
In Table 5.4.3, we deviate from the base case in terms of the service radius (increase to 20
miles) and the number of pickups per trip (increase to 100). Increasing the coverage increases the
number of miles traveled in the local distance, thus increasing the transport cost. But this effect is
not significant on the overall cost and profit margin.
Table 5.4.3: Larger service area within a county (larger radius, more pickups) Parameters
Transport Cost ($/mile) - $2.38 /mile Number of Households (units) - 100
Storage cost ($) - $0.69 Capacity of the vehicle - 3,500 lbs. Cost of recycling ($/US ton) - $120 Percentage of capacity used - 20% Incentive Cost ($ / Household / Month) - $0 Distance negotiated - No Radius of coverage – 20 miles Type of vehicle (for emission cost) – Type 1
Results Total Cost $28,878,865.92
Total Margin $17,563,977.86 Total Emissions Cost $11,287,545.90
Plots
53
5.4.4 Partnering to Share Logistics Cost
Here we consider that we decide on a consolidation-based logistics strategy, and the CPG
company partners with the 3PL player and pays only for the second leg of the transportation. We
consider this scenario in the assumption, that the 3PL provider would make deliveries anyway and
must come back to the warehouse location, and in the process would just pick up the post-consumer
plastic. This shows an expected increase in the margin for the company because of lower logistics
cost. This also reduces the emissions cost as borne by the sponsoring entity, because we only
consider one leg of the journey and hold the assumption that the Leg 1 of the journey would be
completed anyway by the 3PL provider.
Table 5.4.4: Partnering to share logistics cost Parameters
Transport Cost ($/mile) - $2.38 /mile Number of Households (units) - 50 Storage cost ($) - $0.69 Capacity of the vehicle - 3,500 lbs. Cost of recycling ($/US ton) - $120 Percentage of capacity used - 20% Incentive Cost ($ / Household / Month) - $0 Distance negotiated - Yes
Radius of coverage – 5 miles Type of vehicle (for emission cost) – Type 1 Results
Total Cost $13,529,765.42 Total Margin $32,913,078.36
Total Emissions Cost $1,927,298.44 Plots
54
5.4.5 Impact of The Capacity of the Vehicle Used
Here we consider the impact of the capacity of the vehicle. We deviate from the base case
scenario by changing the capacity of the vehicle to 720,000 lbs. and the emission type to Type 2.
We see that the margin drastically improves and the effect on emissions cost also lowers. This
behavior is attributed to the reduction in the number of trips to collect all the post-consumer plastic.
Table 5.4.5: Impact of capacity of vehicle Parameters
Transport Cost ($/mile) - $2.38 /mile Number of Households (units) - 50 Storage cost ($) - $0.69 Capacity of the vehicle - 720,000 lbs.
Cost of recycling ($/US ton) - $120 Percentage of capacity used - 20% Incentive Cost ($ / Household / Month) - $0 Distance negotiated - No Radius of coverage – 5 miles Type of vehicle (for emission cost) – Type 2
Results Total Cost $5,881,016.13
Total Margin $40,561,827.65 Total Emissions Cost $1,368,050.34
Plots
55
5.4.6 Impact of Percentage of the Vehicle Capacity Used in Type 1 Vehicle
In this scenario we consider 5% of the capacity used for the standard delivery van instead
of 20% as in the base case scenario. This results in a loss for the CPG company, as there are
multiple trips required to pick up the post-consumer plastic.
Table 5.4.6.1: Impact of percentage of the vehicle capacity used in Type 1 vehicle Parameters
Transport Cost ($/mile) - $1.70 /mile Number of Households (units) - 50 Storage cost ($) - $0.69 Capacity of the vehicle - 3,500 lbs. Cost of recycling ($/US ton) - $120 Percentage of capacity used - 5%
Incentive Cost ($ / Household / Month) - $0 Distance negotiated - No Radius of coverage – 5 miles Type of vehicle (for emission cost) – Type 1
Results Total Cost $98,205,102.36
Total Margin $(51,762,258.58) Total Emissions Cost $11,274,605.89
Plots
56
In the following scenario we increase the capacity to 40% to see how the margin and the
emissions cost change. It is important to understand that increasing the capacity of the vehicle
reduces the number of trips required thereby reducing the overall transportation cost and thus
improving the margin. It also reduces the emissions and thereby the emissions cost.
Table 5.4.6.2: Impact of percentage of the vehicle capacity used in Type 1 vehicle Parameters
Transport Cost ($/mile) - $2.38 /mile Number of Households (units) - 50 Storage cost ($) - $0.69 Capacity of the vehicle – 3,500 lbs. Cost of recycling ($/US ton) - $120 Percentage of capacity used - 40%
Incentive Cost ($ / Household / Month) - $0 Distance negotiated - No Radius of coverage – 5 miles Type of vehicle (for emission cost) – Type 1
Results Total Cost $17,323,232.97
Total Margin $29,119,610.81 Total Emissions Cost $11,303,078.51
Plots
57
5.4.7 Impact of the Percentage of Vehicle Capacity Used on Type 2 Vehicle
We now conduct a sensitivity analysis by changing the percentage of capacity used for the
Type 2 vehicle (capacity ~ 720,000 lbs.). We use two scenarios for the percentage used as 5% and
40%. The results for 5% capacity used can be seen in Table 5.4.7.1 and the results for 40% capacity
used can be seen in Table 5.4.7.2.
Table 5.4.7.1: Impact of percentage of the vehicle capacity used in Type 2 vehicle Parameters
Transport Cost ($/mile) - $2.38 /mile Number of Households (units) - 50 Storage cost ($) - $0.69 Capacity of the vehicle – 720,000 lbs.
Cost of recycling ($/US ton) - $120 Percentage of capacity used - 5%
Incentive Cost ($ / Household / Month) - $0 Distance negotiated - No Radius of coverage – 5 miles Type of vehicle (for emission cost) – Type 2
Results Total Cost $ 6,218,023.92
Total Margin $ 40,224,819.86 Total Emissions Cost $ 1,133,117.68
Plots
58
Table 5.4.7.2: Impact of percentage of the vehicle capacity used in Type 2 vehicle Parameters
Transport Cost ($/mile) - $2.38 /mile Number of Households (units) - 50 Storage cost ($) - $0.69 Capacity of the vehicle - 720,000 lbs.
Cost of recycling ($/US ton) - $120 Percentage of capacity used - 40%
Incentive Cost ($ / Household / Month) - $0 Distance negotiated - No Radius of coverage – 5 miles Type of vehicle (for emission cost) – Type 2
Results Total Cost $ 5,824,848.17
Total Margin $ 40,617,995.61 Total Emissions Cost $ 1,681,293.88
Plots
59
Based on the results in Tables 5.4.7.1 and 5.4.7.2 we can see that there is not much of a
difference in the cost and margin in these two scenarios. The difference is primarily attributed to
the local distance covered during the pick-up and not the line-haul distance. This suggests that
when using both 5% and 40% of the capacity for a large vehicle the number of trips is
approximately the same for constant demand, which in this case is the consumed weight of the
post-consumer plastic. This compared with Section 5.4.6, where we do the analysis based on the
Type 1 vehicle, is very different because for the smaller vehicle the number of trips is greatly
increased.
5.4.8 Impact on Greenhouse Gas Emissions
In this scenario analysis we focus on the greenhouse gas emissions for the two primarily
types of vehicles we use in this model, viz. Type 1 and Type 2. These two vehicles vary in capacity
60
and in their carbon intensity (CO2e / ton-mile). Carbon intensity for Type 1 is 780 and for Type 2
it is 73. We calculate the emissions based on the GLEC framework (Greene & Lewis, 2019)
Table 5.4.8.1: Emissions in Type 1 vehicle Parameters
Transport Cost ($/mile) - $2.38 /mile Number of Households (units) - 50 Storage cost ($) - $0.69 Capacity of the vehicle - 3,500 lbs.
Cost of recycling ($/US ton) - $120 Percentage of capacity used - 20% Incentive Cost ($ / Household / Month) - $0 Distance negotiated - No Radius of coverage – 5 miles Type of vehicle (for emission cost) – Type 1
Results Total Cost $28,877,785.74
Total Margin $17,565,058.04 Total Emissions Cost $11,286,808.44
Total Emission 418029.94 Metric Tons CO2e Plot
Table 5.4.8.2: Emissions in Type 2 vehicle Parameters
Transport Cost ($/mile) - $2.38 /mile Number of Households (units) - 50 Storage cost ($) - $0.69 Capacity of the vehicle - 720,000 lbs.
Cost of recycling ($/US ton) - $120 Percentage of capacity used - 20% Incentive Cost ($ / Household / Month) - $0 Distance negotiated - No Radius of coverage – 5 miles Type of vehicle (for emission cost) – Type 2
Results
CO2
61
Total Cost $5,881,016.13 Total Margin $40,561,827.65
Total Emissions Cost $1,368,050.34 Total Emission 50668.53 Metric Tons CO2e
Plot
Based on this sensitivity analysis we can clearly see that the emissions are lower if we use
the larger vehicle.
5.4.9 Impact of Incentives
In this research we have always considered that 100% of the post-consumer plastic sold is
collected using the take-back process described in Chapter 2. This is primarily due to lack of data
describing the relationships between the actual post-consumer collections and other tangible
incentives that drive this behavior and enable the actor company to perform this take-back process.
In this scenario, we will test our model cost with a $100 per household per month incentive to get
better collection rates from that household. The incentives considered in this case are static
incentives and results in a flat increase in the overall cost. This investment primarily incentivizes
better collection rates, and create awareness about the company’s environmental prerogatives.
CO2
62
Table 5.4.9: Impact of customer incentives Parameters
Transport Cost ($/mile) - $2.38 /mile Number of Households (units) - 50 Storage cost ($) - $0.69 Capacity of the vehicle - 3,500 lbs. Cost of recycling ($/US ton) - $120 Percentage of capacity used - 20% Incentive Cost ($ / Household / Month) - $100 Distance negotiated - No Radius of coverage – 5 miles Type of vehicle (for emission cost) – Type 1
Results Total Cost $ 32,837,785.74
Total Margin $ 13,605,058.04 Total Emissions Cost $ 11,286,808.44
Plots
63
6. Discussion
After understanding the results from Initial Data Analysis, Optimized Routes, Margin and
Cost Analysis, and Scenario-based Sensitivity Analysis, we now discuss the results. This chapter
will cover three sections: (1) A sensitivity parameter-based analysis of the result, (2) Stakeholder
incentive analysis, and (3) Recommendations.
This project investigates the potential for an e-commerce-based, reverse logistics
mechanism to improve take-back of used plastic packaging from the end consumers to the source
or value generation point, such that the cost of logistics will be less than the combined value of
cost of production of plastic, value generated from optimized take-back, value of the intangible
brand value improvement, and overall cost of responsibility to manage waste management.
The current model and contribution are focused on CPG industry data extrapolated from
the data provided by a CPG company in the North American region. While conducting this
research we have focused on the New England region to execute the model and analyze the results.
This research, however, is not bounded by any company periphery or geographic region and can
be extended to other organizations and regions with finite additional varying parameters.
6.1 Sensitivity Parameter-Based Analysis of the Results
Based on the sensitivity analysis conducted in Section 5.4, we understood how the system
behaves while we use different parameters to effect change in the system cost and behavior. We
comment on some of the observations from the sensitivity analysis, which are significant.
6.1.1 Impact of the choice of type of vehicle on logistics cost
64
The Type 1 vehicle considered as the base case in the sensitivity analysis performed is the
generic van (capacity ≤ 3.5 tons). In our case we considered the vehicle of capacity 3,500 lbs.
which is a generic urban logistics vehicle used by prominent logistics players. In our analysis we
found that if we use the Type 1 vehicle, we end up making a greater number of trips compared to
when we use the Type 2 vehicle – a larger truck (capacity ≥ 70 ton). We make a smaller number
of trips when we use the larger truck, in which case the difference in the profit margin is
approximately $23 million. The bigger the trucks the higher the profits.
6.1.2 Impact of percentage capacity utilization of the vehicle used on logistics cost
Similar to the discussion in Section 6.1.1, we studied the change in logistics cost based on
percentage capacity utilization of the vehicle used on logistics cost. We understand that reducing
the percentage capacity will increase the number of trips. This can be grasped in a similar vein as
reducing the overall capacity of the vehicle. In our sensitivity analysis we test our data with a 5%
capacity utilization of the Type 1 vehicle. This simulation makes the profit margin go negative,
generating a significant loss of $51.76 million. Increasing the amount of plastic, a vehicle takes
back increases profits.
6.1.3 Impact of vehicle type on emissions
Emissions is an important consideration while analyzing a network design approach to a
problem. This is more important in this case as we try to find a solution to an environmentally
detrimental problem, primality to improve the quality of the environment. We need to be careful
that, while solving one problem we are not introducing new ones. The case of emissions is similar
in this context. While we are trying to reduce the global plastic waste, we do not want to increase
the systems impact on the climate as a result.
65
We assess this in our sensitivity analysis through analyzing the carbon intensity of the type
of vehicle used. We find that, based on the average emissions data from the GLEC Framework,
the Type 1 vans are approximately nine times as harmful as the Type 2 vehicles. Based on the
study we conclude it is better to use the Type 2 vehicle for all the line-haul type distances. Higher
capacity vehicles cause lower emissions.
6.1.4 Impact of paying for the complete distance rather than Leg 2
In this solution design, we propose that the 3PL providers pick up the post-consumer plastic
after doing deliveries from the same households or apartments. This is an optimal approach and
removes additional round trips from the overall process, thereby increasing efficiency and reducing
cost. However, if the CPG company pays for both the legs of the transportation journey of the
vehicle from the County to the Amazon warehouse (Leg 1) and from the Amazon warehouse to
the MRF (Leg 2), then the CPG company is essentially paying for half of the delivery trip for
Amazon. The CPG company should not do that and instead negotiate to pay for only the second
leg of the journey.
We performed sensitivity analysis for this mode of operation in Section 5.4.4 and found
that if the CPG company pays only for the Leg 2 section of the journey, then the profit margin
almost doubles to $32 million.
6.2 Stakeholder Incentive Analysis
Every system has stakeholders. Stakeholders are the major actors of an underlying system
who interact with the system either actively or passively. Stakeholders are the most important piece
for the operability and effectiveness of the system from both a process and a cost standpoint
66
(Veiga, 2013). In this project, the stakeholders are the consumer of the products, the third-party
logistics (3PL) provider, and the MRFs.
We briefly describe the stakeholders and their actions as the following:
6.2.1 Consumers
The end consumer of CPG products is the start of the reverse supply chain of plastic waste.
The consumer uses the product and creates the waste, which needs to be routed and recycled.
Consumers are the actors who take part in the take-back process by facilitating the pickup of the
plastic waste by a 3PL provider or mail them. But before this, an aware consumer might already
be recycling properly, without contaminating the packaging or throwing it out in the trash. Why
would the consumer take on this extra task – what incentivizes him or her?
1. Consumer is motivated to act as a responsible citizen to reduce plastic pollution through
superior waste management. Both incentives and information can be a driver of recycling
behavior for the consumers, however, dissemination of information seems to have longer-
term effects than incentives (both positive in terms of rewards, and negative in terms of
punishment payment) (Iyer & Kashyap, 2007).
2. Consumer coupon incentives had been useful in the 1990s for the recycling of the
aluminum cans (Allen, Davis, & Soskin, 1993) and similar structure is also used in Finland
for plastic and has been received with high enthusiasm (Abila & Kantola, 2019).
3. It has also been seen that the payment by weight for recycling yields better results than a
flat fee for municipal solid waste (MSW) (Thøgersen, 2003). However, this is not the case
in the current scenario for MSW since the plastic-import ban by China because recycling
67
is broken in the US (Corkery, 2019). But if an entity chooses to facilitate the take-back
process for post-consumer plastic this incentive scheme will become effective again.
4. An empirical study has estimated the willingness to pay (WTP) for the consumers for
recycling initiatives to be $2.29 (after adjusting bias). This estimation is based on a survey
conducted in a southeastern US neighborhood. This study also indicates the long-term
ineffectiveness of the incentive program because it inures the group of people considered
in the study. (Koford, Blomquist, Hardesty, & Troske, 2012)
Based on these studies, we can conclude that the CPG companies can employ a low-value
incentive structure for a short period of time, to encourage consumer participation in the plastic
waste take-back process, while raising awareness at the same time. Thereby the incentives can
wane out with time, as the process solidifies.
6.2.2 3rd Party Logistics Providers
In an e-commerce-based take-back mechanism, the major cost of the overall process is
logistics. The total logistics cost includes transportation cost, sorting and handling cost, and the
collection cost. Based on the model, the consumer either hands in the waste to an e-commerce 3PL
provider or mails it directly to the warehouse of the 3PL provider. This incurs cost at the 3PL
provider end, and the same must be incurred by the CPG company to enable the take-back process.
Here the incentives are the cost of the processes charged by the 3PL provider. However, there is a
possibility that the CPG companies could convince the 3PL providers to reduce the cost over time
because of the gains through goodwill by aiding the recycling of plastic (Srinivasan & Singh, 2010)
and also reduce the logistics cost by committing to long-term contracts (Sink, Langley, & Gibson,
1996).
68
6.2.3 Material Recovery Facility (MRF)
The final, and the culminating stakeholder in the system is the MRF, who has the final
responsibility of preparing, processing and recycling the collected solid plastic waste from the
consumer at the end of the supply chain (Pressley, Levis, Damgaard, Barlaz, & DeCarolis, 2015).
The several processes as mentioned in Section 4.5 incur a variety of costs, which include the fixed
cost of the systems, labor cost and variable operations cost (Chang & Wang, 1995). This cost has
to be borne by the party which wants control of the recycled plastic at the end of the recycling
process, which in this case is the CPG company.
The implications of the stakeholder analysis are primarily twofold: (1) Improving the
collection percentage of the post-consumer plastic, and (2) maximizing the addressable market to
reap maximum returns.
In our model we have considered a static cost of incentives which are not tied with the
percentage of the post-consumer plastic collected. Also, in our model, this take-back process
assumes 100% conversion of all the plastic sold as CPG products into post-consumer plastic and
a 100% of that is collected using the take-back process. This assumption, however, is untrue in
some circumstances, such as where the consumer is not a responsible recycler in this system. That
is to say that she doesn’t give back the used plastic through this process but produces additional
waste. In such situations, incentives have been proven helpful. The incentive can be designed as a
dollar amount to each household who is responsible actors in this system and can enable higher
collection rates of post-consumer plastic. Other actors such as the 3PL providers are important
stakeholders and several decisions are dependent on pricing and other commercial terms.
In conclusion to this analysis of stakeholder incentives, we understand that even though
initially the consumers can be enticed with monetary benefits, it is not required in the longer term.
69
This can be explained – as the consumers get habitually inclined to participating in the take-back
process, also we find that there is already a WTP towards a plastic-free society. Of the next two
costs, viz. the 3PL provider costs and MRF costs: The 3PL costs are more dominant, compared to
the costs incurred at the MRFs (Chang, Davila, Dyson, & Brown, 2005). Therefore, the major cost
component or stakeholder incentive for the CPG company is the cost incurred for the services
procured from the 3PL provider, which can further be reduced through long-term contracts (Sink
et al., 1996) and bulk deliveries (Goldsby & Closs, 2000).
6.3 Recommendation
Based on the discussion in Section 6.1 and 6.2 we are now well placed to make strategic
recommendations to the CPG company.
6.3.1 Best Case Scenario
To maximize the profits when using e-commerce-based reverse logistics channel approach
to take-back of post-consumer plastic, the CPG company should use a Type 2 vehicle at 20%
capacity utilization. This recommendation is based upon the results as seen in Section 5.4.5. The
profit margin is slightly lower than utilizing 40%, but realistically it is difficult to acquire 40% of
a vehicle for plastic-waste collection purposes in a real-world reverse logistics process. This
selection also minimizes the emissions.
6.3.2 Worst Case Scenario
To maximize the profits when using e-commerce-based reverse logistics channel approach
to take-back of post-consumer plastic, CPG company should not use of Type 1 vehicles at 5%
capacity utilization, as this will generate heavy losses of approximately $50 million annually. This
70
recommendation is based upon the results as seen in Section 5.4.6. This process is the most harmful
for the environment in terms of emissions with an estimated 418,000 ton-CO2.
Based on these scenarios analyzed in this project, we recommend breaking down the
transportation based on the vehicle types for the different legs of the network, by using Type 1
vehicle in the first leg and the Type 2 vehicle in the second leg.
6.4 Contribution
The major contributions of this thesis are primarily at the juncture where sustainability
meets process improvements. At a time, when recycling is of utmost importance, there is a need
for a process change in the collection of post-consumer plastic to increase the amount of plastic
collected for recycling, while creating an economically, socially, and environmentally feasible
process for the CPG companies. This thesis contributes to the existing literature by defining both
quantitatively and qualitatively the use of existing e-commerce reverse logistics channels to take
back post-consumer plastics.
In doing so quantitatively, we contribute by proposing a MILP-based network design
model (as described in Section 4.4) to optimize the cost of operations (as described in Section 4.1
(8)), and by designing a total cost equation (as described in Section 4.5) to calculate the profit
margin of the company facilitating this process, thereby assessing its economic feasibility.
Using the quantitative analysis and corresponding data as a foundation for scenario-based
sensitivity analysis (as described in Section 5.4), this thesis contributes qualitatively by designing
a tool to assess the economic and environmental feasibility of the process at various scenarios by
changing different parameters affecting the system overall.
71
7. Conclusion
This research vividly explains how a model leveraging an e-commerce-based reverse
logistics channel can facilitate an economically viable take-back process for post-consumer plastic
while keeping emissions minimal and adding to the value of the CPG companies. We started to
dive into the problem after stating a research and proving the viability of that idea. After
performing a rigorous literature review it was clear that e-commerce reverse logistics channel has
never been used to take back post-consumer plastic directly from households while simultaneously
performing e-commerce deliveries. This notion as substantiated by the literature review was very
valuable while performing this research, because we know that the e-commerce reverse logistics
network is already set up around us and can be easily leveraged to facilitate this process.
We made some assumptions along the way, which were:
(1) We considered 100% collection rate of post-consumer plastic which enabled us to consider the
normalized sales data directly as our working demand for the post-consumer plastic.
(2) We assumed that this process can access a percentage of space in the vehicles used for
transportation, and that the 3PL providers will make room for post-consumer plastic and prevent
any contamination of new products.
(3) We also made a similar assumption about warehouses freeing up a percentage of their capacity
to consolidate the post-consumer plastic from different counties.
(4) Even though we are aware the other regions might look very different in terms of demographics
and geographic placement of MRFs and Amazon warehouses, it is important to note that this model
will correctly point out the feasibility of effecting this take-back process. Here we assume that this
process will render a positive outcome as we see in the New England region.
72
Further research is suggested, particularly around the following items:
(1) Understanding the relationship of the stakeholder initiatives with the post-consumer plastic
collection, trucking price negotiation, symbiotic relationships with competitors.
(2) Defining metrics to quantitatively understand the stakeholder incentives as a variable in the
model.
(3) Understanding the environmental value in greater depth from an economic standpoint. This
entails an understanding of the monetary implications of brand value improvement, cost of
mitigating plastic pollution, attainability of sustainable initiatives, and using new processes as a
step towards contributing to the United Nations Sustainable Development goals.
(4) Expanding this study to other types of wastes and assessing if an e-commerce based closed-
loop supply chain would be feasible for take-back of other types of post-consumer wastes such as
glass, cardboard, and aluminum using the model and techniques proposed in this thesis.
73
References
Abila, B., & Kantola, J. (2019). The perceived role of financial incentives in promoting waste recycling—
empirical evidence from Finland. Recycling. https://doi.org/10.3390/recycling4010004
Allen, J., Davis, D., & Soskin, M. (1993). Using coupon incentives in recycling aluminum : A market app.
The Journal of Consumer Affairs. https://doi.org/10.1111/j.1745-6606.1993.tb00750.x
Alliance To End Plastic Waste. (n.d.). Retrieved May 4, 2020, from https://endplasticwaste.org/
Atasu, A., Toktay, L. B., & Van Wassenhove, L. N. (2013). How Collection Cost Structure Drives a
Manufacturer’s Reverse Channel Choice. Production and Operations Management, 22(5), n/a-n/a.
https://doi.org/10.1111/j.1937-5956.2012.01426.x
Atasu, A., & Van Wassenhove, L. (2011). Getting to Grips With Take-Back Laws: What’s Yours Is Mine.
IESE Insight, (8), 29–35. https://doi.org/10.15581/002.art-1892
Chang, N. Bin, Davila, E., Dyson, B., & Brown, R. (2005). Optimal design for sustainable development of a
material recovery facility in a fast-growing urban setting. Waste Management.
https://doi.org/10.1016/j.wasman.2004.12.017
Chang, N. Bin, & Wang, S. F. (1995). The development of material recovery facilities in the United States:
status and cost structure analysis. Resources, Conservation and Recycling.
https://doi.org/10.1016/0921-3449(94)00041-3
Chow, C. F., So, W. M. W., Cheung, T. Y., & Yeung, S. K. D. (2017). Plastic waste problem and education
for plastic waste management. In Emerging Practices in Scholarship of Learning and Teaching in a
Digital Era (pp. 125–140). https://doi.org/10.1007/978-981-10-3344-5_8
Corkery, M. (2019). As Costs Skyrocket, More U.S. Cities Stop Recycling - The New York Times. Retrieved
74
April 30, 2020, from https://www.nytimes.com/2019/03/16/business/local-recycling-costs.html
Elliott, P., Shaddick, G., Kleinschmidt, I., Jolley, D., Walls, P., Beresford, J., & Grundy, C. (1996). Cancer
incidence near municipal solid waste incinerators in Great Britain. British Journal of Cancer, 73(5),
702–710. https://doi.org/10.1038/bjc.1996.122
Fråne, A., Stenmarck, Å., Gíslason, S., Lyng, K., & Løkke, S. (2014). Collection & recycling of plastic waste:
Improvements in existing collection and recycling systems in the Nordic countries. Retrieved from
https://books.google.com/books?hl=en&lr=&id=HeHEAwAAQBAJ&oi=fnd&pg=PA71&dq=related:ZI
A9fyUM9qkJ:scholar.google.com/&ots=RkjKbkl_hY&sig=dLOP_esV4WL_e4kD51Uf3QOHqUI
General Secretariat of the European Parliament and of the Council. (2019). Directive of the European
Parliament and of the Council on the reduction of the impact of certain plastic products on the
environment. Retrieved from https://data.consilium.europa.eu/doc/document/ST-5483-2019-
INIT/en/pdf
Geyer, R., Jambeck, J. R., & Law, K. L. (2017). Production, use, and fate of all plastics ever made. Science
Advances. https://doi.org/10.1126/sciadv.1700782
Goldsby, T. J., & Closs, D. J. (2000). Using activity-based costing to reengineer the reverse logistics
channel. International Journal of Physical Distribution and Logistics Management.
https://doi.org/10.1108/09600030010372621
Govindan, K., Palaniappan, M., Zhu, Q., & Kannan, D. (2012). Analysis of third party reverse logistics
provider using interpretive structural modeling. International Journal of Production Economics,
140(1), 204–211. https://doi.org/10.1016/j.ijpe.2012.01.043
Greene, S., & Lewis, A. (2019). Global Logistics Emissions Council Framework for Logistics Emissions
Accounting and Reporting Smart Freight Centre.
75
Hegberg, B. A., Hallenbeck, W. H., & Brenniman, G. R. (1993). Plastics recycling rates. Resources,
Conservation and Recycling, 9(1–2), 89–107. https://doi.org/10.1016/0921-3449(93)90035-E
Hopewell, J., Dvorak, R., & Kosior, E. (2009, July 27). Plastics recycling: Challenges and opportunities.
Philosophical Transactions of the Royal Society B: Biological Sciences, Vol. 364, pp. 2115–2126.
https://doi.org/10.1098/rstb.2008.0311
Iyer, E. S., & Kashyap, R. K. (2007). Consumer recycling: role of incentives, information, and social class.
Journal of Consumer Behaviour. https://doi.org/10.1002/cb.206
Jambeck, J. R., Geyer, R., Wilcox, C., Siegler, T. R., Perryman, M., Andrady, A., … Law, K. L. (2015). Plastic
waste inputs from land into the ocean. Science. https://doi.org/10.1126/science.1260352
Jeswani, H. K., & Azapagic, A. (2016). Assessing the environmental sustainability of energy recovery from
municipal solid waste in the UK. Waste Management, 50, 346–363.
https://doi.org/10.1016/j.wasman.2016.02.010
John Hocevar. (2020). Circular Claims Fall Flat: Comprehensive U.S. Survey of Plastics Recyclability -
Greenpeace.
Katz, C. (2019). Piling Up: How China’s Ban on Importing Waste Has Stalled Global Recycling - Yale E360.
Retrieved May 4, 2020, from https://e360.yale.edu/features/piling-up-how-chinas-ban-on-
importing-waste-has-stalled-global-recycling
Klausner, M., & Hendrickson, C. T. (2000). Reverse-logistics strategy for product take-back. Interfaces.
https://doi.org/10.1287/inte.30.3.156.11657
Koford, B. C., Blomquist, G. C., Hardesty, D. M., & Troske, K. R. (2012). Estimating consumer willingness
to supply and willingness to pay for curbside recycling. Land Economics.
https://doi.org/10.3368/le.88.4.745
76
Kokkinaki, A. I., Dekker, R., De Koster, M. B. M., Pappis, C., & Verbeke, W. (2002). E-business models for
reverse logistics: Contributions and challenges. Proceedings - International Conference on
Information Technology: Coding and Computing, ITCC 2002.
https://doi.org/10.1109/ITCC.2002.1000434
Locations of Amazon Fulfillment Centers in USA - Forest Shipping. (n.d.). Retrieved May 4, 2020, from
https://forestshipping.com/amazon-fulfillment-usa
Narancic, T., & O’Connor, K. E. (2019, February 1). Plastic waste as a global challenge: Are biodegradable
plastics the answer to the plastic waste problem? Microbiology (United Kingdom), Vol. 165, pp.
129–137. https://doi.org/10.1099/mic.0.000749
New Hampshire Code of Administrative Rules. (2017).
Phillips, V. (2018). Reducing plastic bags in the City of Boston | Boston.gov. Retrieved April 12, 2020,
from https://www.boston.gov/departments/environment/reducing-plastic-bags-city-boston
Pressley, P. N., Levis, J. W., Damgaard, A., Barlaz, M. A., & DeCarolis, J. F. (2015). Analysis of material
recovery facilities for use in life-cycle assessment. Waste Management.
https://doi.org/10.1016/j.wasman.2014.09.012
Residential MRFs - The Recycling Partnership. (n.d.). Retrieved May 4, 2020, from
https://recyclingpartnership.org/residential-mrfs/
Sink, H. L., Langley, C. J., & Gibson, B. J. (1996). Buyer observations of the US third-party logistics market.
International Journal of Physical Distribution & Logistics Management.
https://doi.org/10.1108/09600039610115009
Srinivasan, S., & Singh, R. K. (2010). The persistence of green goodwill. Environment, Development and
Sustainability. https://doi.org/10.1007/s10668-009-9226-z
77
Tammemagi, H. Y. (1999). Search for a Sustainable Future. Retrieved from Oxford University Press
website: http://bibfra.me/vocab/lite/
The World Bank. (2020). Carbon Pricing Dashboard | Up-to-date overview of carbon pricing initiatives.
Retrieved April 29, 2020, from https://carbonpricingdashboard.worldbank.org/map_data
Thøgersen, J. (2003). Monetary Incentives and Recycling: Behavioural and Psychological Reactions to a
Performance-Dependent Garbage Fee. Journal of Consumer Policy.
https://doi.org/10.1023/a:1023633320485
Tonglet, M., Phillips, P. S., & Bates, M. P. (2004). Determining the drivers for householder pro-
environmental behaviour: Waste minimisation compared to recycling. Resources, Conservation and
Recycling, 42(1), 27–48. https://doi.org/10.1016/j.resconrec.2004.02.001
Udall, S. T. (2020). Break Free From Plastic Pollution Act of 2020.
US EPA, O. (n.d.). Plastics: Material-Specific Data.
Veiga, M. M. (2013). Analysis of efficiency of waste reverse logistics for recycling. Waste Management
and Research. https://doi.org/10.1177/0734242X13499812
Verma, R., Vinoda, K., Papireddy, M., & Gowda, A. (2016). Toxic Pollutants from Plastic Waste- A Review.
Procedia Environmental Sciences, 35, 701–708. https://doi.org/10.1016/j.proenv.2016.07.069
Whitehouse, S. (2019). Save Our Seas 2.0 Act Advances Through Final Senate Committee. Retrieved April
12, 2020, from https://www.whitehouse.senate.gov/news/release/save-our-seas-20-act-advances-
through-final-senate-committee
78
Appendix This appendix lists tables depicting some of the relevant, directly applicable data analyses which enabled us to perform the
scenario-based sensitivity analysis. The results from the scenario-based analysis were then used to understand the pros and cons of the
process and make effective recommendations, substantiating the claims made in this research. The lists included in this Appendix are:
(1) Amount of Plastic Generated by County – This table shows the plastic consumption by county. This value is obtained by multiplying
the normalizing parameter with the CPG sales information. The normalizing parameter is calculated as mentioned in Section 4.2.4; (2)
County ID Mapping – This table shows the IDs assigned to the counties to achieve consistency in analysis; (3) MRF ID Mapping - This
table shows the IDs assigned to the MRFs to achieve consistency in analysis; (4) Amazon Warehouse ID Mapping - This table shows
the IDs assigned to the Amazon Warehouses to achieve consistency in analysis; (5) Cost, Price and Margin Calculations – This table
focuses on actual calculation of cost, total price of virgin plastic, and final calculation of the profit margin as described in Section 4.5;
and (6) Distance Matrix – The distance matrix depicts the output from running the optimization model as described in Section 4.4. This
table contains both direct distances between a county and an MRF, and the aggregated distance where a consolidator is involved.
A. Amount of Plastic Generated by County
States and Counties
Normalizing Parameter
Total (Metric Tons)
PET (Metric Tons)
PP (Metric Tons)
HDPE (Metric Tons)
LDPE (Metric Tons)
Connecticut 0.019652 10519.91474 4333.255 5989.648 192.9864 4.024888 Fairfield County 0.008191 4384.792497 1806.139 2496.538 80.43843 1.677609 Hartford County 0.004165 2229.707805 918.4383 1269.513 40.90369 0.85308
79
Litchfield County 0.000846 453.050442 186.6159 257.9501 8.311149 0.173336 Middlesex County 0.000804 430.22451 177.2137 244.9538 7.892411 0.164603 New Haven County 0.003495 1871.089882 770.7201 1065.329 34.32489 0.715873 New London County 0.001137 608.891171 250.8082 346.68 11.17003 0.23296 Tolland County 0.000615 329.067737 135.5462 187.3589 6.036703 0.1259 Windham County 0.000398 213.023454 87.74643 121.2876 3.907887 0.081502 Maine 0.004709 2520.868176 1038.37 1435.289 46.24499 0.964477 Androscoggin County 0.000322 172.205456 70.9331 98.04739 3.159086 0.065885 Aroostook County 0.000199 106.579298 43.90105 60.68229 1.955183 0.040777 Cumberland County 0.001321 706.895936 291.1773 402.4803 12.96791 0.270456 Franklin County 0.000083 44.389575 18.2845 25.27377 0.814321 0.016983 Hancock County 0.000202 108.183979 44.56203 61.59594 1.984621 0.041391 Kennebec County 0.000395 211.325893 87.04719 120.3211 3.876745 0.080853 Knox County 0.00015 80.222212 33.04431 45.67555 1.471666 0.030693 Lincoln County 0.000126 67.370964 27.75075 38.35852 1.235911 0.025776 Oxford County 0.000158 84.31591 34.73054 48.00635 1.546764 0.032259 Penobscot County 0.000457 244.697155 100.7931 139.3215 4.488936 0.09362 Piscataquis County 0.000046 24.838871 10.23137 14.14233 0.455666 0.009503 Sagadahoc County 0.000132 70.426296 29.00927 40.09812 1.291961 0.026945 Somerset County 0.000139 74.300367 30.60504 42.30387 1.36303 0.028427 Waldo County 0.000117 62.530157 25.75678 35.60235 1.147107 0.023924 Washington County 0.000092 48.9947 20.18139 27.89576 0.898801 0.018745 York County 0.000772 413.519708 170.3328 235.4428 7.585963 0.158211 Massachusetts 0.035595 19054.98927 7848.935 10849.2 349.5612 7.290382 Barnstable County 0.001148 614.428628 253.0891 349.8328 11.27161 0.235079 Berkshire County 0.000514 274.955104 113.2567 156.5492 5.044014 0.105197 Bristol County 0.002195 1174.963086 483.9787 668.9803 21.55454 0.449537 Dukes County 0.000117 62.591298 25.78196 35.63716 1.148229 0.023947 Essex County 0.00384 2055.56677 846.7079 1170.363 37.7091 0.786454 Franklin County 0.000279 149.464324 61.56581 85.09943 2.741903 0.057185 Hampden County 0.001732 927.333961 381.9779 527.9895 17.01182 0.354795 Hampshire County 0.000615 329.265189 135.6275 187.4714 6.040326 0.125976
80
Middlesex County 0.009734 5210.968774 2146.449 2966.932 95.59452 1.993701 Nantucket County 0.000101 54.133212 22.298 30.82144 0.993066 0.020711 Norfolk County 0.004683 2507.020481 1032.666 1427.404 45.99095 0.959179 Plymouth County 0.002516 1347.059773 554.867 766.9658 24.71163 0.515381 Suffolk County 0.004713 2522.997363 1039.247 1436.501 46.28405 0.965291 Worcester County 0.003407 1823.996239 751.3218 1038.516 33.46097 0.697856 New Hampshire 0.005982 3202.106303 1318.979 1823.16 58.74221 1.225116 Belknap County 0.000268 143.560208 59.13384 81.73785 2.633593 0.054926 Carroll County 0.000207 110.747032 45.61778 63.05524 2.031639 0.042371 Cheshire County 0.000286 153.097069 63.06217 87.16778 2.808545 0.058574 Coos County 0.000098 52.458063 21.60799 29.86767 0.962336 0.02007 Grafton County 0.00039 208.92286 86.05735 118.9529 3.832662 0.079933 Hillsborough County 0.001827 977.874737 402.7961 556.7655 17.93898 0.374132 Merrimack County 0.00063 337.119998 138.863 191.9436 6.184421 0.128981 Rockingham County 0.001642 878.977939 362.0595 500.4574 16.12473 0.336294 Strafford County 0.000473 253.367957 104.3647 144.2583 4.648001 0.096938 Sullivan County 0.00016 85.893859 35.38051 48.90477 1.575711 0.032863 Rhode Island 0.004172 2233.49857 919.9997 1271.671 40.97323 0.85453 Bristol County 0.000276 147.653162 60.81977 84.06822 2.708677 0.056492 Kent County 0.000703 376.497393 155.0829 214.3636 6.906794 0.144047 Newport County 0.000408 218.33292 89.93345 124.3106 4.005288 0.083534 Providence County 0.002203 1179.228504 485.7357 671.4089 21.63279 0.451169 Washington County 0.000582 311.80169 128.4341 177.5283 5.71996 0.119294 Vermont 0.002441 1306.684967 538.2362 743.9779 23.97096 0.499934 Addison County 0.000138 73.761227 30.38296 41.9969 1.35314 0.028221 Bennington County 0.00014 74.725284 30.78007 42.5458 1.370825 0.02859 Caledonia County 0.00009 48.12501 19.82316 27.40059 0.882847 0.018412 Chittenden County 0.000721 386.067384 159.0249 219.8124 7.082354 0.147708 Essex County 0.000016 8.665834 3.569541 4.934004 0.158974 0.003316 Franklin County 0.000164 87.997921 36.2472 50.10275 1.61431 0.033668 Grand Isle County 0.00003 16.187836 6.667927 9.216752 0.296964 0.006193 Lamoille County 0.000098 52.550046 21.64587 29.92004 0.964024 0.020105
81
Orange County 0.000098 52.659606 21.691 29.98242 0.966033 0.020147 Orleans County 0.000084 45.098095 18.57634 25.67718 0.827318 0.017254 Rutland County 0.000221 118.254291 48.71009 67.3296 2.169359 0.045244 Washington County 0.000252 134.914469 55.57258 76.81528 2.474988 0.051618 Windham County 0.000157 83.86575 34.54512 47.75004 1.538506 0.032087 Windsor County 0.000231 123.754337 50.97561 70.46112 2.270257 0.047348
B. County ID Mapping
State County ID County State County ID County CT CTY_CT_1 Fairfield County ME CTY_ME_12 Androscoggin County CT CTY_CT_2 Hartford County ME CTY_ME_13 Kennebec County CT CTY_CT_3 Litchfield County ME CTY_ME_14 Lincoln County CT CTY_CT_4 Middlesex County ME CTY_ME_15 Oxford County CT CTY_CT_5 New Haven County NH CTY_NH_1 Strafford County CT CTY_CT_6 New London County NH CTY_NH_2 Sullivan County CT CTY_CT_7 Tolland County NH CTY_NH_3 Hillsborough County CT CTY_CT_8 Windham County NH CTY_NH_4 Merrimack County MA CTY_MA_1 Barnstable County NH CTY_NH_6 Rockingham County MA CTY_MA_2 Berkshire County NH CTY_NH_7 Carroll County MA CTY_MA_3 Bristol County NH CTY_NH_8 Cheshire County MA CTY_MA_4 Dukes County NH CTY_NH_9 Coos County MA CTY_MA_5 Essex County NH CTY_NH_10 Grafton County MA CTY_MA_6 Franklin County NH CTY_NH_11 Belknap County MA CTY_MA_7 Hampden County RI CTY_RI_1 Newport County MA CTY_MA_8 Hampshire County RI CTY_RI_2 Providence County MA CTY_MA_10 Middlesex County RI CTY_RI_3 Washington County MA CTY_MA_11 Nantucket County RI CTY_RI_4 Bristol County MA CTY_MA_12 Norfolk County RI CTY_RI_5 Kent County
82
MA CTY_MA_13 Plymouth County VT CTY_VT_1 Windsor County MA CTY_MA_14 Suffolk County VT CTY_VT_2 Orleans County MA CTY_MA_15 Worcester County VT CTY_VT_3 Windham County ME CTY_ME_1 Waldo County VT CTY_VT_4 Franklin County ME CTY_ME_2 Washington County VT CTY_VT_5 Grand Isle County ME CTY_ME_3 York County VT CTY_VT_6 Lamoille County ME CTY_ME_4 Penobscot County VT CTY_VT_7 Bennington County ME CTY_ME_5 Piscataquis County VT CTY_VT_8 Caledonia County ME CTY_ME_6 Sagadahoc County VT CTY_VT_9 Chittenden County ME CTY_ME_7 Somerset County VT CTY_VT_10 Essex County ME CTY_ME_8 Aroostook County VT CTY_VT_11 Addison County ME CTY_ME_9 Cumberland County VT CTY_VT_12 Rutland County ME CTY_ME_10 Franklin County VT CTY_VT_13 Washington County ME CTY_ME_11 Hancock County VT CTY_VT_14 Orange County
C. MRF ID Mapping
State MRF ID MRF Address CT MRF_CT_1 61 Crescent St, Stamford, CT 06906, USA
CT MRF_CT_2 100 3rd St, Bridgeport, CT 06607, USA
CT MRF_CT_3 90 Oliver Terrace, Shelton, CT 06484, USA
CT MRF_CT_4 283 White St, Danbury, CT 06810, USA
CT MRF_CT_5 174 Edgewood Ave, New Britain, CT 06051, USA
CT MRF_CT_6 1680 W Main St, Windham, CT 06280, USA
CT MRF_CT_7 143 Murphy Rd, Hartford, CT 06114, USA
CT MRF_CT_8 143 Murphy Rd, Hartford, CT 06114, USA MA MRF_MA_1 45 Kings Hwy, West Wareham, MA 02576, USA
MA MRF_MA_2 70 Battles St, Brockton, MA 02302, USA
83
MA MRF_MA_3 Main/Cumberland, Springfield, MA 01107, USA
MA MRF_MA_4 13 Robbie Rd, Avon, MA 02322, USA
MA MRF_MA_5 1 Hardscrabble Rd, Auburn, MA 01501, USA MA MRF_MA_6 30 Hopkinton Rd, Westborough, MA 01581, USA
MA MRF_MA_7 40 Bunker Hill Industrial Park, Boston, MA 02129, USA
MA MRF_MA_8 73 Newbury St, Peabody, MA 01960, USA
MA MRF_MA_9 31 High St, North Billerica, MA 01862, USA
ME MRF_ME_1 2300 Congress St, Portland, ME 04102, USA
ME MRF_ME_2 424 River Rd, Lewiston, ME 04240, USA
NH MRF_NH_1 12 Brown Rd, Newport, NH 03773, USA
RI MRF_RI_1 98 Taylor Rd, Johnston, RI 02919, USA VT MRF_VT_1 127 Dorr Dr, Rutland, VT 05701, USA
VT MRF_VT_2 Avenue D, Williston, VT 05495, USA
D. Amazon Warehouse ID Mapping
State Amazon ID Amazon Fulfilment Center Address NH AMZ_1 10 State St, Nashua, NH 03063, USA MA AMZ_2 1180 Innovation Way, Fall River, MA 02720, USA CT AMZ_3 29 Research Pkwy, Meriden, CT 06450, USA CT AMZ_4 409 Washington Ave, North Haven, CT 06473, USA CT AMZ_5 801 Day Hill Rd, Windsor, CT 06095, USA MA AMZ_6 Amazon Fulfillment Center, 1000 Technology Center Dr, Stoughton, MA 02072, USA
84
E. Cost, Price and Margin Calculation
Costs Price Profit
CTY_ID Total (miles)
Plastic Waste (Lbs.)
Logistics Cost ($)
Recycling Cost (includes
sorting) ($)
Storage Cost ($)
Incentive Cost ($) Total P_virgin ($) Margin ($) Emissions
g-CO2 Emission Cost ($)
CTY_CT_1 12.14 9666801.23 $ 798,286.16 $ 580,008.07 $ - $ 6,000.00 $ 5,254,290.00 $ 3,869,995.77 1622887724 43817.97 CTY_CT_2 25.04 4915658.42 $ 837,109.40 $ 294,939.51 $ 56,284.55 $ 6,000.00 $ 2,671,855.38 $ 1,477,521.93 3509934137 94768.22 CTY_CT_3 57.59 998804.07 $ 391,270.80 $ 59,928.24 $ 11,436.36 $ 6,000.00 $ 542,889.64 $ 74,254.23 3773193254 101876.22 CTY_CT_4 35.21 948481.56 $ 227,172.07 $ 56,908.89 $ 10,860.16 $ 6,000.00 $ 515,537.36 $ 214,596.23 1339323127 36161.72 CTY_CT_5 30.95 4125042.18 $ 868,234.10 $ 247,502.53 $ 47,231.95 $ 6,000.00 $ 2,242,124.08 $ 1,073,155.50 4499554359 121487.97 CTY_CT_6 64.56 1342373.65 $ 589,468.59 $ 80,542.42 $ 15,370.25 $ 6,000.00 $ 729,633.34 $ 38,252.08 6372732973 172063.79 CTY_CT_7 12.45 725469.31 $ 61,483.71 $ 43,528.16 $ - $ 6,000.00 $ 394,321.36 $ 283,309.48 128217831.6 3461.88 CTY_CT_8 58.65 469635.77 $ 187,450.66 $ 28,178.15 $ 5,377.35 $ 6,000.00 $ 255,265.67 $ 28,259.51 1841078799 49709.13 CTY_MA_1 90.65 1354581.64 $ 835,193.11 $ 81,274.90 $ 15,510.03 $ 6,000.00 $ 736,268.87 $ (201,709.17) 12678154382 342310.17
CTY_MA_10 32.14 11488205.98 $ 2,510,995.76 $ 689,292.36 $ 131,540.57 $ 6,000.00 $ 6,244,295.75 $ 2,906,467.06 13514199858 364883.40 CTY_MA_11 124.62 119343.16 $ 101,436.02 $ 7,160.59 $ 1,366.49 $ 6,000.00 $ 64,867.74 $ (51,095.35) 2116904647 57156.43 CTY_MA_12 8.11 5527027.49 $ 304,932.22 $ 331,621.65 $ 63,284.76 $ 6,000.00 $ 3,004,158.73 $ 2,298,320.10 414077078.7 11180.08 CTY_MA_13 9.91 2969754.92 $ 200,160.42 $ 178,185.30 $ - $ 6,000.00 $ 1,614,179.62 $ 1,229,833.91 332060735.9 8965.64 CTY_MA_14 20.41 5562250.45 $ 772,085.74 $ 333,735.03 $ 63,688.06 $ 6,000.00 $ 3,023,303.80 $ 1,847,794.97 2638662029 71243.87 CTY_MA_15 23.80 4021218.59 $ 650,834.49 $ 241,273.12 $ - $ 6,000.00 $ 2,185,691.85 $ 1,287,584.24 2593514694 70024.90 CTY_MA_2 79.32 606171.52 $ 327,135.48 $ 36,370.29 $ 6,940.70 $ 6,000.00 $ 329,478.27 $ (46,968.19) 4345136415 117318.68 CTY_MA_3 38.00 2590347.12 $ 669,484.38 $ 155,420.83 $ 29,659.61 $ 6,000.00 $ 1,407,956.43 $ 547,391.61 4260258173 115026.97 CTY_MA_3 106.60 137990.03 $ 100,289.24 $ 8,279.40 $ 1,579.99 $ 6,000.00 $ 75,003.06 $ (41,145.58) 1790380891 48340.28 CTY_MA_4 18.18 4531743.61 $ 560,343.62 $ 271,904.62 $ - $ 6,000.00 $ 2,463,182.45 $ 1,624,934.21 1705804383 46056.72 CTY_MA_5 67.30 329512.04 $ 150,968.43 $ 19,770.72 $ 3,772.93 $ 6,000.00 $ 179,102.87 $ (1,409.21) 1701470538 45939.70 CTY_MA_5 5.20 2044419.00 $ 72,302.60 $ 122,665.14 $ - $ 6,000.00 $ 1,111,222.84 $ 910,255.10 62910347.34 1698.58 CTY_MA_6 55.66 725904.62 $ 274,915.22 $ 43,554.28 $ 8,311.65 $ 6,000.00 $ 394,557.96 $ 61,776.82 2562670547 69192.10 CTY_MA_7 228.49 137855.23 $ 214,738.54 $ 8,271.31 $ 1,578.45 $ 6,000.00 $ 74,929.79 $ (155,658.51) 8216737732 221851.92 CTY_MA_8 251.55 97862.14 $ 168,003.99 $ 5,871.73 $ 1,120.53 $ 6,000.00 $ 53,191.96 $ (127,804.29) 7077412774 191090.14 CTY_ME_1 297.64 238504.56 $ 483,433.35 $ 14,310.27 $ 2,730.89 $ 6,000.00 $ 129,636.69 $ (376,837.82) 24096343588 650601.28
85
CTY_ME_10 11.21 379647.59 $ 28,978.05 $ 22,778.86 $ - $ 6,000.00 $ 206,353.53 $ 148,596.62 54396349.38 1468.70 CTY_ME_11 199.73 465893.29 $ 633,245.09 $ 27,953.60 $ 5,334.50 $ 6,000.00 $ 253,231.49 $ (419,301.71) 21180724745 571879.57 CTY_ME_12 193.30 148527.37 $ 195,701.14 $ 8,911.64 $ 1,700.65 $ 6,000.00 $ 80,730.52 $ (131,582.90) 6335193456 171050.22 CTY_ME_13 195.35 185884.54 $ 247,398.34 $ 11,153.07 $ 2,128.39 $ 6,000.00 $ 101,035.62 $ (165,644.18) 8093529064 218525.28 CTY_ME_14 330.69 108014.70 $ 243,682.37 $ 6,480.88 $ 1,236.77 $ 6,000.00 $ 58,710.27 $ (198,689.76) 13494819724 364360.13 CTY_ME_15 113.44 911653.82 $ 703,511.74 $ 54,699.23 $ 10,438.48 $ 6,000.00 $ 495,520.02 $ (279,129.43) 13364396486 360838.71 CTY_ME_2 304.14 539464.24 $ 1,116,439.16 $ 32,367.85 $ 6,176.89 $ 6,000.00 $ 293,220.22 $ (867,763.69) 56864282339 1535335.62 CTY_ME_3 311.50 54760.27 $ 116,741.32 $ 3,285.62 $ 627.01 $ 6,000.00 $ 29,764.38 $ (96,889.56) 6089866366 164426.39 CTY_ME_4 179.84 155263.22 $ 190,306.21 $ 9,315.79 $ 1,777.77 $ 6,000.00 $ 84,391.72 $ (123,008.06) 5731397429 154747.73 CTY_ME_5 310.50 163804.08 $ 346,601.26 $ 9,828.24 $ 1,875.57 $ 6,000.00 $ 89,034.01 $ (275,271.05) 18022639003 486611.25 CTY_ME_6 414.13 234966.85 $ 662,681.64 $ 14,098.01 $ 2,690.38 $ 6,000.00 $ 127,713.81 $ (557,756.23) 45959035813 1240893.97 CTY_ME_7 14.96 1558436.92 $ 158,604.00 $ 93,506.22 $ - $ 6,000.00 $ 847,072.30 $ 588,962.08 397314940.6 10727.50 CTY_ME_8 74.75 558580.07 $ 284,117.87 $ 33,514.80 $ 6,395.77 $ 6,000.00 $ 303,610.43 $ (26,418.02) 3556574714 96027.52 CTY_ME_9 114.06 460595.52 $ 357,530.09 $ 27,635.73 $ 5,273.84 $ 6,000.00 $ 250,351.94 $ (146,087.73) 6829312259 184391.43 CTY_NH_1 82.76 316495.71 $ 178,321.45 $ 18,989.74 $ 3,623.89 $ 6,000.00 $ 172,027.97 $ (34,907.11) 2471440771 66728.90
CTY_NH_10 3.02 189363.32 $ 3,908.31 $ 11,361.80 $ - $ 6,000.00 $ 102,926.48 $ 81,656.37 1978167.82 53.41 CTY_NH_11 32.90 2155842.20 $ 482,402.42 $ 129,350.53 $ 24,684.51 $ 6,000.00 $ 1,171,785.77 $ 529,348.31 2657634054 71756.12 CTY_NH_2 69.36 743221.49 $ 350,703.61 $ 44,593.29 $ 8,509.93 $ 6,000.00 $ 403,970.37 $ (5,836.45) 4073326344 109979.81 CTY_NH_3 67.71 1937812.34 $ 892,374.69 $ 116,268.74 $ 22,188.05 $ 6,000.00 $ 1,053,277.89 $ 16,446.41 10118060949 273187.65 CTY_NH_4 118.54 244155.12 $ 197,091.47 $ 14,649.31 $ 2,795.59 $ 6,000.00 $ 132,708.00 $ (87,828.37) 3912418282 105635.29 CTY_NH_6 67.75 337520.86 $ 155,668.77 $ 20,251.25 $ 3,864.63 $ 6,000.00 $ 183,455.98 $ (2,328.67) 1766189339 47687.11 CTY_NH_7 193.57 115650.09 $ 152,696.56 $ 6,939.01 $ 1,324.20 $ 6,000.00 $ 62,860.42 $ (104,099.35) 4949919637 133647.83 CTY_NH_8 52.61 481341.12 $ 172,327.92 $ 28,880.47 $ 5,511.38 $ 6,000.00 $ 261,628.00 $ 48,908.23 1518174480 40990.71 CTY_NH_9 12.96 2599750.74 $ 229,124.03 $ 155,985.04 $ - $ 6,000.00 $ 1,413,067.68 $ 1,021,958.60 497081634.9 13421.20 CTY_RI_1 84.73 687404.24 $ 396,283.41 $ 41,244.25 $ 7,870.82 $ 6,000.00 $ 373,631.48 $ (77,767.00) 5623090026 151823.43 CTY_RI_2 21.97 325519.11 $ 48,686.37 $ 19,531.15 $ - $ 6,000.00 $ 176,932.55 $ 102,715.03 179093871.8 4835.53 CTY_RI_3 21.92 830033.68 $ 123,777.49 $ 49,802.02 $ - $ 6,000.00 $ 451,156.24 $ 271,576.73 454290571.9 12265.85 CTY_RI_5 124.43 272831.29 $ 231,156.99 $ 16,369.88 $ 3,123.93 $ 6,000.00 $ 148,294.63 $ (108,356.17) 4816847056 130054.87 CTY_VT_1 186.62 19104.87 $ 24,698.65 $ 1,146.29 $ 218.75 $ 6,000.00 $ 10,384.26 $ (21,679.44) 771881398.6 20840.80
CTY_VT_11 188.47 162615.48 $ 208,859.69 $ 9,756.93 $ 1,861.96 $ 6,000.00 $ 88,387.96 $ (138,090.62) 6591985085 177983.60 CTY_VT_12 4.16 260705.78 $ 7,394.90 $ 15,642.35 $ - $ 6,000.00 $ 141,703.93 $ 112,666.68 5151217.456 139.08 CTY_VT_13 168.44 297435.14 $ 341,081.51 $ 17,846.11 $ 3,405.65 $ 6,000.00 $ 161,667.80 $ (206,665.47) 9620975711 259766.34 CTY_VT_14 137.32 116094.42 $ 108,738.79 $ 6,965.67 $ 1,329.29 $ 6,000.00 $ 63,101.92 $ (59,931.82) 2500570116 67515.39 CTY_VT_2 203.13 99424.16 $ 137,825.28 $ 5,965.45 $ 1,138.41 $ 6,000.00 $ 54,040.98 $ (96,888.17) 4688519655 126590.03 CTY_VT_3 114.11 184892.11 $ 143,750.36 $ 11,093.53 $ 2,117.02 $ 6,000.00 $ 100,496.20 $ (62,464.71) 2747078681 74171.12 CTY_VT_4 239.82 194001.98 $ 316,949.05 $ 11,640.12 $ 2,221.33 $ 6,000.00 $ 105,447.77 $ (231,362.73) 12729104795 343685.83 CTY_VT_5 242.67 35688.03 $ 59,476.44 $ 2,141.28 $ 408.63 $ 6,000.00 $ 19,397.86 $ (48,628.49) 2417052257 65260.41
86
CTY_VT_6 206.26 115852.88 $ 162,990.58 $ 6,951.17 $ 1,326.52 $ 6,000.00 $ 62,970.64 $ (114,297.64) 5629988954 152009.70 CTY_VT_7 150.44 164740.86 $ 168,893.45 $ 9,884.45 $ 1,886.29 $ 6,000.00 $ 89,543.19 $ (97,121.00) 4255011139 114885.30 CTY_VT_8 176.06 106097.36 $ 127,444.10 $ 6,365.84 $ 1,214.82 $ 6,000.00 $ 57,668.12 $ (83,356.64) 3757494933 101452.36 CTY_VT_9 10.58 851131.88 $ 61,255.55 $ 51,067.91 $ - $ 6,000.00 $ 462,623.94 $ 344,300.49 108479314 2928.94
87
F. Distances Matrix
CTY_ID
MRF_ID Miles Final
for Cost
Total Distance
(including Leg 1)
CTY_ID
AMZ_ID MRF_ID Miles CTY_CT_1 MRF_CT_4 12.14 12.14 12.14 CTY_CT_1 AMZ_4 MRF_CT_3 59.21 CTY_CT_2 MRF_CT_8 14.63 13.19 25.04 CTY_CT_2 AMZ_5 MRF_CT_7 25.04 CTY_CT_3 MRF_CT_4 34.15 13.91 57.59 CTY_CT_3 AMZ_3 MRF_CT_5 57.59 CTY_CT_4 MRF_CT_5 24.15 13.91 35.21 CTY_CT_4 AMZ_3 MRF_CT_5 35.21 CTY_CT_5 MRF_CT_3 22.08 21.26 30.95 CTY_CT_5 AMZ_4 MRF_CT_3 30.95 CTY_CT_6 MRF_CT_6 20.02 13.19 64.56 CTY_CT_6 AMZ_5 MRF_CT_7 64.56 CTY_CT_7 MRF_CT_6 12.45 12.45 12.45 CTY_CT_7 AMZ_5 MRF_CT_7 37.09 CTY_CT_8 MRF_CT_6 16.26 13.19 58.65 CTY_CT_8 AMZ_5 MRF_CT_7 58.65 CTY_MA_1 MRF_MA_1 33.35 27.58 90.65 CTY_MA_1 AMZ_2 MRF_MA_2 90.65 CTY_MA_10 MRF_MA_9 15.41 1.63 32.14 CTY_MA_10 AMZ_6 MRF_MA_4 32.14 CTY_MA_11 MRF_MA_1 67.28 27.58 124.62 CTY_MA_11 AMZ_2 MRF_MA_2 124.62 CTY_MA_12 MRF_MA_4 7.51 1.63 8.11 CTY_MA_12 AMZ_6 MRF_MA_4 8.11 CTY_MA_13 MRF_MA_1 9.91 9.91 9.91 CTY_MA_13 AMZ_2 MRF_MA_2 67.58 CTY_MA_14 MRF_MA_7 3.23 1.63 20.41 CTY_MA_14 AMZ_6 MRF_MA_4 20.41 CTY_MA_15 MRF_MA_6 23.80 23.80 23.80 CTY_MA_15 AMZ_1 MRF_MA_9 80.51 CTY_MA_2 MRF_MA_3 46.54 13.19 79.32 CTY_MA_2 AMZ_5 MRF_CT_7 79.32 CTY_MA_3 MRF_RI_1 28.36 27.58 38.00 CTY_MA_3 AMZ_2 MRF_MA_2 38.00 CTY_MA_4 MRF_MA_1 49.30 27.58 106.60 CTY_MA_4 AMZ_2 MRF_MA_2 106.60 CTY_MA_5 MRF_MA_8 18.18 18.18 18.18 CTY_MA_5 AMZ_1 MRF_MA_9 69.84 CTY_MA_6 MRF_MA_3 34.53 13.19 67.30 CTY_MA_6 AMZ_5 MRF_CT_7 67.30 CTY_MA_7 MRF_MA_3 5.20 5.20 5.20 CTY_MA_7 AMZ_5 MRF_CT_7 39.03 CTY_MA_8 MRF_MA_3 22.89 13.19 55.66 CTY_MA_8 AMZ_5 MRF_CT_7 55.66 CTY_ME_1 MRF_ME_2 71.17 24.10 228.49 CTY_ME_1 AMZ_1 MRF_MA_9 228.49 CTY_ME_10 MRF_ME_2 91.03 24.10 251.55 CTY_ME_10 AMZ_1 MRF_MA_9 251.55 CTY_ME_11 MRF_ME_2 140.32 24.10 297.64 CTY_ME_11 AMZ_1 MRF_MA_9 297.64 CTY_ME_12 MRF_ME_2 11.21 11.21 11.21 CTY_ME_12 AMZ_1 MRF_MA_9 172.44 CTY_ME_13 MRF_ME_2 42.41 24.10 199.73 CTY_ME_13 AMZ_1 MRF_MA_9 199.73 CTY_ME_14 MRF_ME_2 50.45 24.10 193.30 CTY_ME_14 AMZ_1 MRF_MA_9 193.30 CTY_ME_15 MRF_ME_2 52.87 24.10 195.35 CTY_ME_15 AMZ_1 MRF_MA_9 195.35 CTY_ME_2 MRF_ME_2 174.21 24.10 330.69 CTY_ME_2 AMZ_1 MRF_MA_9 330.69 CTY_ME_3 MRF_ME_1 27.41 24.10 113.44 CTY_ME_3 AMZ_1 MRF_MA_9 113.44 CTY_ME_4 MRF_ME_2 146.83 24.10 304.14 CTY_ME_4 AMZ_1 MRF_MA_9 304.14 CTY_ME_5 MRF_ME_2 155.40 24.10 311.50 CTY_ME_5 AMZ_1 MRF_MA_9 311.50 CTY_ME_6 MRF_ME_2 37.45 24.10 179.84 CTY_ME_6 AMZ_1 MRF_MA_9 179.84 CTY_ME_7 MRF_ME_2 136.75 24.10 310.50 CTY_ME_7 AMZ_1 MRF_MA_9 310.50 CTY_ME_8 MRF_ME_2 256.82 24.10 414.13 CTY_ME_8 AMZ_1 MRF_MA_9 414.13 CTY_ME_9 MRF_ME_1 14.96 14.96 14.96 CTY_ME_9 AMZ_1 MRF_MA_9 145.58
88
CTY_NH_1 MRF_ME_1 53.67 24.10 74.75 CTY_NH_1 AMZ_1 MRF_MA_9 74.75 CTY_NH_10 MRF_NH_1 61.36 24.10 114.06 CTY_NH_10 AMZ_1 MRF_MA_9 114.06 CTY_NH_11 MRF_NH_1 52.30 24.10 82.76 CTY_NH_11 AMZ_1 MRF_MA_9 82.76 CTY_NH_2 MRF_NH_1 3.02 3.02 3.02 CTY_NH_2 AMZ_1 MRF_MA_9 99.76 CTY_NH_3 MRF_MA_9 32.61 24.10 32.90 CTY_NH_3 AMZ_1 MRF_MA_9 32.90 CTY_NH_4 MRF_NH_1 34.83 24.10 69.36 CTY_NH_4 AMZ_1 MRF_MA_9 69.36 CTY_NH_6 MRF_MA_9 36.60 24.10 67.71 CTY_NH_6 AMZ_1 MRF_MA_9 67.71 CTY_NH_7 MRF_ME_1 56.83 24.10 118.54 CTY_NH_7 AMZ_1 MRF_MA_9 118.54 CTY_NH_8 MRF_NH_1 37.70 24.10 67.75 CTY_NH_8 AMZ_1 MRF_MA_9 67.75 CTY_NH_9 MRF_ME_2 96.86 24.10 193.57 CTY_NH_9 AMZ_1 MRF_MA_9 193.57 CTY_RI_1 MRF_RI_1 35.06 27.58 52.61 CTY_RI_1 AMZ_2 MRF_MA_2 52.61 CTY_RI_2 MRF_RI_1 12.96 12.96 12.96 CTY_RI_2 AMZ_2 MRF_MA_2 59.31 CTY_RI_3 MRF_RI_1 34.39 27.58 84.73 CTY_RI_3 AMZ_2 MRF_MA_2 84.73 CTY_RI_4 MRF_RI_1 21.97 21.97 21.97 CTY_RI_4 AMZ_2 MRF_MA_2 42.36 CTY_RI_5 MRF_RI_1 21.92 21.92 21.92 CTY_RI_5 AMZ_2 MRF_MA_2 74.90 CTY_VT_1 MRF_NH_1 27.70 24.10 124.43 CTY_VT_1 AMZ_1 MRF_MA_9 124.43 CTY_VT_10 MRF_VT_2 95.28 24.10 186.62 CTY_VT_10 AMZ_1 MRF_MA_9 186.62 CTY_VT_11 MRF_VT_2 27.10 24.10 188.47 CTY_VT_11 AMZ_1 MRF_MA_9 188.47 CTY_VT_12 MRF_VT_1 4.16 4.16 4.16 CTY_VT_12 AMZ_1 MRF_MA_9 160.90 CTY_VT_13 MRF_VT_2 45.29 24.10 168.44 CTY_VT_13 AMZ_1 MRF_MA_9 168.44 CTY_VT_14 MRF_VT_1 43.79 24.10 137.32 CTY_VT_14 AMZ_1 MRF_MA_9 137.32 CTY_VT_2 MRF_VT_2 67.38 24.10 203.13 CTY_VT_2 AMZ_1 MRF_MA_9 203.13 CTY_VT_3 MRF_VT_1 55.14 24.10 114.11 CTY_VT_3 AMZ_1 MRF_MA_9 114.11 CTY_VT_4 MRF_VT_2 39.52 24.10 239.82 CTY_VT_4 AMZ_1 MRF_MA_9 239.82 CTY_VT_5 MRF_VT_2 42.38 24.10 242.67 CTY_VT_5 AMZ_1 MRF_MA_9 242.67 CTY_VT_6 MRF_VT_2 37.66 24.10 206.26 CTY_VT_6 AMZ_1 MRF_MA_9 206.26 CTY_VT_7 MRF_VT_1 56.68 24.10 150.44 CTY_VT_7 AMZ_1 MRF_MA_9 150.44 CTY_VT_8 MRF_VT_2 81.83 24.10 176.06 CTY_VT_8 AMZ_1 MRF_MA_9 176.06 CTY_VT_9 MRF_VT_2 10.58 10.58 10.58 CTY_VT_9 AMZ_1 MRF_MA_9 196.15
89
[The End]